Uploaded by Hector Gonzalez


Friedemann Tischmeyer
Internal Mixing
How to create a professional mix on your computer – a systematic approach
Create punchy, powerful, clear,
and three-dimensional mixes on Mac
and PC-based digital audio workstations.
Tischmeyer Publishing
Second, revised edition 2008
Publisher and Author:
Friedemann Tischmeyer
Brian Smith
Cover Design:
aim Werbeagentur Hamburg, Michael Prahl
Mott Jordan www.mottjordan.com
Friedemann Tischmeyer and Gregoire Vanoli
Leina Gonzalez Baird
Omid Bürgin, Namin Nooman
Media Print GmbH, printed in Germany
optimal media production GmbH, manufactured in Germany
©2008 by Tischmeyer Publishing GmbH Germany
All rights reserved.
No part of this book may be reproduced, stored in a retrieval system, or transmitted in any form or by
any means, mechanical, electronic, photocopying, recording, or otherwise, without the prior written
permission of the publisher.
ISBN: 978-3-9811217-1-1
The contents of this book and the accompanying DVD have been written and produced with the
greatest possible care. However, nobody is perfect and it is possible errors have occurred. The
publisher cannot accept any responsibility for damages resulting from erroneous information, nor
from the use of the software included on the DVD.
Tischmeyer Publishing cannot provide software support for included demo software. Please contact
the respective software manufacturer.
The Internet addresses in this book are given for informative purposes only – the publisher is not
responsible for their content.
Dear Reader,
successful mix.
ELCOME ABOARD! We are about to embark on a voyage
through the entire mixing process. This book will give you easyto-understand suggestions on how to systematically carry out a
Very often people ask if computer-based production can provide the same punch
and especially the same degree of space that a production made with classical
tools and the best outboard equipment. Step by step, I will explain the necessary
tools and techniques for professional mixing with computers. We will look at
how mixing was done with tape-based recording technology, and will examine
the supposed advantages and disadvantages of these production methods while
applying this to the world of digital production.
On the way, we will discover a treasure of inspiration for improving our current
working methods.
Beginners, DJs, audio engineers, musicians, producers, and audio engineering
students will all be able to use the ideas and suggestions in this book for their
Here’s a suggestion for all those whose highest priority is the best possible conservation of their ideas and songs: In order to make progress in mixing, please
plan on finishing your projects within a predetermined period of time. This is the
only way that you will be able to look back and see how your work methods and
your hearing have developed. On the other hand, if you let yourself be a slave to
the possibilities of total recall incorporated in today’s digital audio workstations
(DAW) by never really finishing your projects, then it will be much more difficult
to see any progress in your mixing techniques.
The software and hardware names in this book are property of the companies that own them and are
protected under local and international copyright laws.
The Sound Examples contained on the DVD are only intended for use in exercises or as examples.
Any other use is expressly forbidden.
AM A TRAINED guitar and bass player and have learned everything I
know about audio engineering on my own. As both a musician and as an
engineer, I have been very lucky to have worked with a variety of experienced engineers; I have used these experiences and opportunities to further increase my knowledge. Later I had a large 48-track studio with a Studer 2” tape
recorder and a Trident analog mixing desk/console, along with a large quantity of
outboard equipment. During this time, I gained quite a bit of experience working
with tape – quite a bit different from the predominately individual or mini-team
working methods that are characteristic of today’s production methods using
home computers. Before the digital revolution could wipe out my studio, I was
still able to sell off my equipment to concentrate fully on the new computer-based
techniques in a smaller studio, spending more time working on my own projects.
I constantly tried to achieve results of a quality equal to that attained using the
familiar analog techniques. In the beginning, it was not easy to attain the same
quality with digital workstations as analog technology. Therefore, I began to work
with manufacturers and developers of software-based workstations and plug-ins,
and always went to the limits of what was technically feasible.
Please do not use any excerpts as samples for productions as their use is not allowed without express permission from the copyright owners.
The knowledge contained in my books therefore stems from pure practical user
experience, from one user to another. Wherever appropriate, it is supported by
extensive research and studies.
Enjoy yourself and have fun applying the following techniques to your work!
Friedemann Tischmeyer
Like my mastering book “Audio Mastering with PC Workstations,” this book has
“organically” grown from the many workshops I have given in my studio and at
various educational institutions. Workshops given to small groups are ideal for
the mutual exchange of information and experiences. They helped me to finetune pedagogical concepts and to tailor to the participants’ needs – to your needs.
This was the foundation of both the book and the DVD series I created based
on the books. Theoretical issues, which were inappropriate for the DVDs, along
with all information about quickly evolving matters such as plug-in descriptions,
are reserved for the book, which will be revised in a cycle of approximately two
to three years. On the other hand, the tutorial DVDs focus on practical working methods illustrated by numerous audio examples. The book’s accompanying
DVD-ROM contains audio examples, exercises, and demo versions of plug-ins
from many different manufacturers.
I would like to thank the owners of copyrighted material who have allowed me to
reproduce parts of their work for the audio examples. There are also single-track
excerpts that are suitable for exercises concerning compressors and EQs.
EARLY ALL OF THE INFORMATION and suggestions in this
book are cross-platform and are therefore equally valid for both
PC and Mac users; exceptions will be indicated. The same is the
case for mixing strategies with analog mixing consoles and computer-based
systems. Nevertheless, the focus of the book is clearly based on computerbased mixing.
information in this book, you will be able to correctly navigate yourself through
the entire mixing process.
After exploring the individual areas in depth, the section “Workflow Overview”
represents the mixing process as a timeline in order to give you an overview of
the “absolute” process workflow.
For the sake of clarity, screenshots are taken from only one sequencer program.
Because I personally work with Steinberg’s Nuendo and 99% of the features that
concern us also are included in Cubase, I have used screenshots from Nuendo.
To avoid unnecessary complication, I will not discuss other sequencers. Every
professional sequencer or hard disk recording application has the necessary basic functions for our work and it will not be difficult for you to apply the workflows that we describe to your own software environment. My choice is not a
recommendation concerning the very similar qualities of the various sequencers or DAWs available today.
In this book we will concentrate almost entirely on systematically organized
techniques for interpreting and answering artistic requirements. The right
sound and how to obtain it is our main task, provided you already have a clear
idea of the sound aesthetic and the artistic flow you want.
The system involving the complex, intuitive, and creative process of music mixing can be divided into two large categories: absolute processes and relative processes.
Absolute processes refer to recurring steps or rules, which should be carried
out in a specific manner and order. For example it makes sense to begin with
automation after finishing the static mix.
Relative processes also follow rules, but are not dependent on being done in a
strict order. Working with EQs, while independent of the mixing process order, still follows the same rules. Relative process steps can therefore occur at
different points in the mixing process or even more often than once. It will be
self-evident which steps are relative and which are absolute processes. With the
Foreword . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3
Concerning my Background and the Creation of this Book . . . . . . . . . . . . . . 4
Working with this Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6
Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9
Chapter 1: The Three Phases of Classical Production –
A Retrospective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Phase 1 – The Recording Session . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
Improve your Decision-making Capabilities . . . . . . . . . . . . . . . . . . . . . . . 20
Building Confidence in Rhythmic Hearing . . . . . . . . . . . . . . . . . . . . . . . . . 21
A Few Good Tracks are Better than Many Mediocre Ones! . . . . . . . . . . . 21
Phase 2 - The Mixdown . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
How to Achieve a Quality Mix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Mixing Goals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
The Three Dimensions: Aspects/Sub-Aspects . . . . . . . . . . . . . . . . . . . . . . . 24
Delivering the Mix Master to the Mastering Studio . . . . . . . . . . . . . . . . . . 25
Phase 3 - Mastering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
Chapter 2: The Technical Requirements for a Good Sounding
Computer-based Mix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Analog vs. Digital Summing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
The Story of Analog Summing Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Prerequisites for Digital Summing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
Working with Summing Units . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
The Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
Device Particularities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
Test Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
More Bits for More Sound! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
PCM – The Principles of Digital Audio . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Bit Resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
Hard Choices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
What is Truncation? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
UAD-1 Card . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Why is High Bit Depth So Important for Sound? . . . . . . . . . . . . . . . . . . . . 48
PowerCore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
What Happens When a 16-bit File is Saved as a 32-bit File? . . . . . . . . . . . 49
Incompatibility Between UAD-1 and PowerCore? . . . . . . . . . . . . . . . . . 79
Why Record in 32-bit Format When the A/D Converter Only Supports
24-bit Resolution? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Waves. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Summary of a Few Basic Rules Relating to Bit Depth . . . . . . . . . . . . . . . . 50
Offline Processing as an Alternative to DSP-supported Plug-Ins . . . . 82
When Can We Leave the 32-bit Domain? . . . . . . . . . . . . . . . . . . . . . . . . . . 51
Mixing Console Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Tip for Mastering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
In-line Consoles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
Dithering. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Split Consoles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Studio Acoustics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
The Cubase and Nuendo Split Console . . . . . . . . . . . . . . . . . . . . . . . . . . 86
How to Live With Imperfect Acoustics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
Description of the Audio Channel Diagram . . . . . . . . . . . . . . . . . . . . . . 90
Studio Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
Choosing Speakers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
The Basics of Setting up Speakers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
Front-end . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Back-end . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
D/A Converters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
Cable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
Monitor Controller, Monitor Matrix, and Mixing Console . . . . . . . . . 61
Power Amplifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
Digital Monitors – the Purist’s Back-end . . . . . . . . . . . . . . . . . . . . . . . . . 62
Remotes and Controllers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
Using Metering for Monitoring the Dimensions . . . . . . . . . . . . . . . . . . . . 64
Chapter 3: The Systematic Approach –
Clarity and Workflow . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Defining the Start of the Mixing Process . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Converting MIDI to Audio. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
Listen To and Clean Up Tracks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
Using Folder Tracks and Colors for Clarity . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Working with Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
Using Compression in Groups: Save CPU and Increase Punch . . . . . . . . . 101
EQing Groups to Save Resources and Create Space . . . . . . . . . . . . . . . . . . . 102
Handling Groups: What to Control via Groups, and what via the Tracks? . 103
Widening Stereo Basis within Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
Pinguin PG-AM 4.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Workflow Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
Peak and Loudness Measurement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Chapter 4: Dimension 1: Horizontal = Panorama . . . . . . . . . . . . . 107
Goniometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Spectrum Analyzers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
Correlation Meter . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
Pinguin Spectrometer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
Extra DSP Power for Software-based Mixing . . . . . . . . . . . . . . . . . . . . . . . 76
SSL Duende . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
Summary of Panning Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Preventing Phase Cancellation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Phase Interference Mostly Occurs in the Following Cases . . . . . . . . . . . 113
The Masking Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
Stereo Pan Mode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Chapter 5: Dimension 2: The Vertical Dimension =
Frequency Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Dimension 2: 1st Aspect: Using EQs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Frequency Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
Description of the Frequency Bands. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
The Most Important Filter Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
Working with EQs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
First Remove and Then Add! . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
A Good Mix Starts with the Arrangement! . . . . . . . . . . . . . . . . . . . . . . . . 127
Mixing a Tight Bass Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
EQ Basic Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
Dimension 2: 2nd Aspect – Level Measurement . . . . . . . . . . . . . . . . . . . . . . 129
Dimension 2: 3rd Aspect – Compression . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
Common Basic Parameters of Compressors and Their Use . . . . . . . . . . 131
Compressor Types . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
Exercises with Compressors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141
Spatial Exercise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
Chapter 7: Additional Artistic Aspects: Design and the
Accessories Mute and Special FX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Mute . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
Special FX . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168
Chapter 8: Automation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
Chapter 9: Working with Individual Instruments . . . . . . . . . . . . 181
Drums . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Bass Drum . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 181
Snare . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184
Other Aspects of Drums . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 186
Refining Loops with Doubling, Gating, Compression, and Reverb . . . . 188
Bass . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
Vocals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 192
Chapter 6: Dimension 3 = Layering with Reverb and Delay . . 147
Acoustic guitar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
Good Reverb is a Prerequisite for Layering . . . . . . . . . . . . . . . . . . . . . . . . . . 147
Electric Guitar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200
Native and DSP-based Reverb Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Keyboards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201
Using External Reverb Devices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149
Chapter 10: Overview of Recommended Plug-ins. . . . . . . . . . . . 203
Dimension 3 and Aspects for Designing Stage Depth . . . . . . . . . . . . . . . . . 150
Reverb as Send Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
Aspect 1: Pre-delay as Sound Design Component . . . . . . . . . . . . . . . . . . 151
Aspect 2: EQing in Reverb Return as a Design Component . . . . . . . . . . 153
Layering Strategies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Basic Reverb Rules. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154
Basic Delay Rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155
Pre-fader or Post-fader? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156
How Many Reverb Devices Do You Need? Planning Depth Design . . . 157
Delay Instead of Reverb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
Delay in Reverb or Reverb in Delay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
EQs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
Universal Audio Cambridge EQ (UAD-1: optional) . . . . . . . . . . . . . . . . 205
Universal Audio Precision EQ (UAD-1: optional) . . . . . . . . . . . . . . . . . . 206
Universal Audio Pultec & Pultec Pro (UAD-1: optional) . . . . . . . . . . . . 207
Universal Audio CS-1 (channel strip on the UAD-1 card) . . . . . . . . . . . 208
Universal Audio Neve 1073 EQ (UAD-1: optional) . . . . . . . . . . . . . . . . . 209
Universal Audio Neve 1081 EQ (UAD-1: optional) . . . . . . . . . . . . . . . . . 209
SSL Duende EQ and Dynamics Channel Strip (included with Duende) . 210
TC EqSat (included in PowerCore) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
TC Dynamic EQ (PowerCore: optional) . . . . . . . . . . . . . . . . . . . . . . . . . . 212
TC VoiceStrip (included in PowerCore). . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Other Dynamic Processing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238
Sonnox Oxford EQ (optional PowerCore and native) . . . . . . . . . . . . . . . 214
Sonnox Oxford Inflator (PowerCore: optional, native) . . . . . . . . . . . . . . 239
Waves Renaissance EQ (REQ4) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
Waves L2 (native) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 240
Waves Linear Phase EQ (LinEq Broadband) . . . . . . . . . . . . . . . . . . . . . . . 216
Steinberg Loudness Maximizer (native). . . . . . . . . . . . . . . . . . . . . . . . . . . 241
Algorithmix Linear Phase PEQ Orange and Red . . . . . . . . . . . . . . . . . . . 217
Universal Audio Precision Maximizer (UAD-1: optional) . . . . . . . . . . . 242
Nomad PEQ2A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218
Sonnox Oxford Transient Modulator (PowerCore: optional, native) . . 243
Sonalksis SV-517 Mk2. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
SPL Transient Designer (UAD-1: optional) . . . . . . . . . . . . . . . . . . . . . . . . 244
Nuendo Channel EQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
Reverb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
EQ: Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221
Audioease Altiverb 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245
Compressors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
SIR2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 246
Universal Audio 1176 LN (UAD-1: optional) . . . . . . . . . . . . . . . . . . . . . . 222
Cubase RoomWorks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247
Universal Audio Neve 33609 Master bus Compressor/Limiter
(UAD-1: optional) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223
Universal Audio RealVerb Pro (UAD-1: included). . . . . . . . . . . . . . . . . . 248
TC 247C (1176 emulation/included with PowerCore). . . . . . . . . . . . . . . 224
Universal Audio Plate 140 (UAD-1: optional) . . . . . . . . . . . . . . . . . . . . . . 249
Sonnox Oxford Dynamics (optional PowerCore and native) . . . . . . . . . 225
TC ClassicVerb (PowerCore: included) . . . . . . . . . . . . . . . . . . . . . . . . . . . 250
Waves Renaissance Compressor – RComp (native) . . . . . . . . . . . . . . . . . 226
TC MegaReverb (PowerCore: included) . . . . . . . . . . . . . . . . . . . . . . . . . . 251
Waves C1 Compressor/Gate (native) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227
TC DVR2 Digital Vintage Reverb (PowerCore: optional) . . . . . . . . . . . . 252
Nomad Blue Tube Compressor CP2S (native) . . . . . . . . . . . . . . . . . . . . . 228
TC NonLin2 (PowerCore: optional) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 253
Voxengo Marquis Compressor (native) . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
TC VSS3 (PowerCore: optional) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
Sonalksis SV-315 Mk2 (native) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229
Waves IR1 (native & suitable for APA) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
PSP Vintage Warmer (native) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 230
Waves Renaissance Reverberator (native) . . . . . . . . . . . . . . . . . . . . . . . . . 255
Universal Audio LA-2A (UAD-1: optional) . . . . . . . . . . . . . . . . . . . . . . . . 231
Voxengo Pristine Space Convolution Reverb (native) . . . . . . . . . . . . . . . 256
Nomad Blue-Tube Vintage Compressor FA-770 (native) . . . . . . . . . . . . 232
Nomad BlueVerb DRV-2080 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
Magneto (native, internal Cubase and Nuendo) . . . . . . . . . . . . . . . . . . . . 232
Other Plug-ins . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257
SSL Duende EQ and Dynamics Channel Strip (included with Duende) . 233
SSL Duende Stereo Bus Compressor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 234
Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258
Limiters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 235
About the Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260
Universal Audio Precision Limiter (UAD-1: optional) . . . . . . . . . . . . . . 236
Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
TC MD3 Brickwall Limiter (PowerCore: optional, in the MD3-Bundle) 236
Sonnox Oxford Limiter (PowerCore: optional and native) . . . . . . . . . . . 237
Nomad Brick Wall Limiter BW2S (native). . . . . . . . . . . . . . . . . . . . . . . . . 238
Universal Audio DreamVerb (UAD-1: optional) . . . . . . . . . . . . . . . . . . . 249
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263
DVD Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
1) Software & Interesting Manuals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
2) Miscellaneous . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
3) Sound Examples (16-bit files) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275
4) Exercises
Compression Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278
Layering Exercise. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
Loop Editing Tempo: 112bpm/Halftime feel. . . . . . . . . . . . . . . . . . . . . 279
Credits for the Sound Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Credits for the Sound Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
Internet-Links. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 289
SE THE ADVANTAGES OF BOTH analog and digital work methods to achieve better results.This chapter adapts the experiences of
classical mixing production in a tape-based studio to the predominating work methods in today’s modern computer-based studios. It helps to take
a brief look into the past to see what has changed and if some of the traditional
methods can be applied today for improving quality.
Traditionally, music production can be divided into three basic work phases
that overlap when working with digital audio workstations (DAWs): Recording
– Mixing – Mastering.
The producer’s job is to make sure that a large number of tracks are filled with
the most musically relevant content possible. Ideally, producers should not have
any personal relationships with the composer, songwriter, singer, musicians, arranger, and engineer, so that they can remain neutral with regards to the production. They need to have a strong imagination in order to record individual tracks,
which – with the exception of a few details – are ready for the mix.
Good sound is created in front of the mic, not behind it!
When working with analog mixers and analog multitrack tape recorders, wellorganized track planning, as well as recording at consistent levels, was necessary
in order to be able to jump quickly from one song to another during an overdub
session. This way, a separate production, or headphone, or rough mix could easily be made by simply adjusting channel faders, panning and headphone send
This technically driven need for highly disciplined organization is diametrically
opposed to the methods of non-destructive computer-based workstations. The
advantages of a practically unlimited number of tracks, loop recording, and total recall of all parameters are all advantages of modern DAWs, as long as we
are consistent in our use of all these possibilities. For example, event-based level
adjustment (which is possible in most audio sequencers) is a quick and helpful
way of correcting recording level inconsistencies. Despite all of its advantages,
loop recording often leads to mediocre results: “there must be a good take in
there somewhere...” The freedom of non-destructive loop recording can become
a forced labor marathon for sound editors. If you’ve already worked professionally with tape media or can mentally imagine the process of tape-based recording, then you can apply a number of valuable tips to your work with DAWs.
Improve your Decision-making Capabilities
The classic method of working trained the ability to decide whether a take was
good or bad. Because of a limited number of tracks and the resulting necessity of
recording over previous takes, decisions had to constantly be made regarding the
quality of each take. To avoid unnecessary searching and editing before things
got out of control, high standards should also be set for nondestructive editing.
For example, while recording a lead vocal track, the workflow could be as follows:
Open a recording track (for listening to the input signal, recording, and punching), a keeper track A (listening) for the “first choice” and a muted “spare parts”
track B. During loop recording, categorize running the takes in writing as A, B,
and C = trash. Another option is to limit the number of loops to a relatively small
number to be judged mentally as you go.
Right after the recording process, the takes are divided between the
keeper track (A), the “spare parts” track (B), and the trash (C), so that
the recording track is empty for the next take. The keeper track is
for listening and should have exactly the same effect settings (insert
and send) as the recording track. Other than small details like manual
tuning, sibilant and breather editing, etc., the A track should sound
good after the session. This makes things much more enjoyable for
the musician, since he or she can have a very good idea of the results
and can go home with a good feeling about the day’s work. Instead of
being overwhelmed by the idea of editing a mountain of disorderly takes, you can
open the project at any time – even weeks later – and still have immediate access
to the tracks that may need editing. During the editing process, the spare parts
track (B) serves as a reserve for replacing details that you might not have heard
with better
during recording (pronunciation, consonant endings and sibilants, unwanted accents, or noises).
Without systematic pre-sorting, you can be fairly sure that after 60 takes of a
loop recording, on the next day you will not remember whether or not THE
take was the 47th, since after the 30th you were probably already somewhat
dazed. In any case, without pre-sorting, your work is laborious, time-consuming, and certainly not much fun.
But by trusting your decision-making abilities, when you finish editing the track
in question, you can make space on your hard disk by completely emptying the
spare parts track (B). Only when you decide to close off theoretical possibilities
can you move forward unhindered.
Building Confidence in Rhythmic Hearing
Back in the days when tape machines were still common, you had to use your
ears to know if a track grooved or not after overdubbing (for example an e-bass
recorded onto drums). Moving tracks was very time-consuming and only in extreme cases could a MIDI-synchronized sampler help with timing problems.
Nowadays, a graphic interface and therefore the eye, is allowed to judge whether
or not a track grooves well or is rhythmically tight. To allow the ears to do the
work they were meant to do, reduce the visual waveform display to a minimum
or eliminate it completely so that you depend 100% on your hearing. After all,
later we want to listen to the music, not watch it! Once the ears have decided that
a sound comes too early or too late, you can switch the waveform display back on
with a keyboard shortcut and go back to the graphic editing mode. Most DAWs
let you change screen layouts with keyboard shortcuts that can be used for this
way of working. Another option is to simply close your eyes while listening for
rhythmic precision and to trust your ears.
A Few Good Tracks are Better than Many
Mediocre Ones!
Another advantage of the limited number of tracks on tape machines was that
bad content could not be compensated by an overly inflated number of tracks. If
20 tracks cannot transport emotion and do not create a particular sound, then it
is highly unlikely that the solution will be found in more tracks.
Clearly there are many reasons why working in analog studios led to more focused, disciplined work that involved spontaneous and quick decision-making,
along with a great deal of imagination. The convenience – inherent in DAWs
– resulting from the lack of clearly-defined production phases has both advantages and disadvantages. Using Cubase, Logic, Pro Tools, DP, or Sonar becomes a
blessing only if we work more systematically.
The “analog” working methods had the advantage that in the second phase – the
mixdown – a good basic sound had already been supplied; the mixing engineer
had a number of fundamental musical aspects already laid out for him. One point
is as valid today as it was in the days of analog recording: the better the recording
and the “front-end” – mics, pre-amps, compressors, and other equipment used
– the more easily a signal integrates into the mix later on. The popular saying
“we’ll fix it in the mix” should not be synonymous with musically or technically
inferior recording, even when we are tempted by the endless correction possibilities of modern DAWs.
In the past, a 2-inch tape was frequently brought to another studio or sent to
a mixing engineer. The mixing studio was usually outfitted with high-end outboard equipment and mixer automation. Phase 2 began with all tracks being laid
down next to each other without panning and all faders set according to taste, in
order to get a first impression. Masking tape was used to label the tracks and a
long tape roll from each song was hung on walls and doors until the production
was finished. Today, the equivalent process would typically be done by importing an OMF (Open Media Framework) file, where individual .wav or .aif files are
brought into the arrangement without EQs, insert effects, or level and panning
The advantages of division of labor here are important: because the sound “grows”
throughout the entire computer-based music production process, by the time the
mixdown begins, a production is often already halfway finished.
That is not necessarily an advantage!
The last fine-tuning can turn into a rocky path because of awkwardness or a lack
of systematic working methods. It is important to keep questions in mind such
as: what is the reason for this send effect? Which strategy is used for panning?
What is my mixing concept?
Being deeply involved in the production can hamper a fresh and systematic approach to the mixing process. Very few people would go to the trouble of taking
out all plug-ins and setting all levels, pans, and EQs to zero and start everything
from the beginning. A further disadvantage in being both the recording and mix
engineer has to do with something we will mention later: the mix concept. Often,
when creating a mix concept, the mute button is a very important tool. The producer, who has been involved since the very start, has a difficult time muting
tracks that were created as the result of hard recording and editing work. This is
why I recommend creating teams with friends and colleagues, and occasionally
changing the roles of mixing engineer and producer with that of the client; taking
on the role of accepting or rejecting final results while delegating the actual mixing process. Doing this will give you new ideas and teach you a great deal.
The mix is done when the mixing engineer creates a stereo master. In earlier days,
this involved a DAT recorder fed by an Apogee converter with onboard limiting.
Often, an analog limiter – like the Urei 1178 LN – was used to cut out peaks in
order to make the best use out of the 16 bits of the DAT recorder. Now that we
have the luxury of 24 or 32-bit mixdowns (bounce/export/render/apply), you do
not have to dynamically limit the master with a bad native converter. The 32-bit
floating-point files provide enough headroom to eliminate a limiter completely,
since even overs can be handled cleanly. The 24-bit files can be protected from
overs using a brick wall limiter, because the limiter only kicks in when a peak is
detected. Generally, dynamic processing and other level processing steps should
be left to the mastering engineer, who takes over in phase 3.
How to Achieve a Quality Mix
An additional artistic aspect is dramatic form and can be especially achieved by
muting or using special effects.
During the recording stage of the production, many tracks are filled with content
that is to be later sorted during the mixing process. Nothing is more boring than
a song where all instruments are audible from beginning to end. By intelligently
muting individual tracks a song can become interesting. For example, if the vocals are good enough you can even create a cappella passages with this process.
Mixing Goals
The goal of a good mix is a warm, clear, deep, and punchy sound, where all events
The secret – or
skill – necessary
to obtain a
good mix lies
in intelligently
distributing all
events in the
three spatial
width, height,
and depth!
u Ask a like-minded friend or colleague to help by switching roles (client/engineer).
u Give yourself only a limited amount of time. In the past, it was often only possible
to have one night to “sleep on” mix decisions – maybe listening on a different set
of monitors – with the sole possibility of coming in the next morning, before the next production day begins, in order to make a few The mixing
minor changes. Afterwards the patch bay and mixer settings would process is a
be irreversibly changed. Do not be led to collect a pile of unfinished
projects just because of the existence of the total recall functions in continuous
modern DAWs. It is much better to decide to close mixing projects chain of
which can be saved and wiped from the hard disk, so that you can
realize that in a year’s time you have gotten better! Then, if required decisions!
you can mix another version. A mix should be 90% done after four hours of
work. The rest is fine-tuning and takes the largest amount of time (1 to 2 days),
but of course this can vary largely and depends on the person.
u The mixing process consists of a continuous chain of decisions. The ability to
quickly make decisions is crucial for maintaining an efficient work pace.
The Three Dimensions
The Three Dimensions:
1. Width (L/R Panorama):
Basis Widening
2. Height (Frequency Distribution):
EQing/ Level
3. Depth (Front-Back Space)
Reverb & Delay
EQing Reverb & Delay
are clearly defined, or correspond to the genre and sound aesthetic. Critically examine every event – other than a quiet background pad for giving warmth – that
does not have a clearly defined place in the mix to see if it would be better to do
without it. Less is often better!
What conclusions can we make from analog techniques for working with
u When you mix your own song and you are not happy with the results, make
a copy of your project, remove all insert and send effects, and put all panning
to the center. Start right from the beginning with a clear mixing strategy (see
following chapters).
Delivering the Mix Master to the
Mastering Studio
As long as you are working on a native basis – in other words, with the computer
and without the aid of an external analog mixing desk – the mix master should
be delivered on CD-ROM as an undithered 32-bit floating point .wav or .aif file.
If your DAW does not provide for the 32-bit floating point format, use the 24-bit
integer format. Since the 24-bit integer format cannot handle overs (level values
exceeding 0 dB), it is essential that you use good brickwall limiting.
To keep the BLER (Block Error Rate) values of the medium (CD-R or DVD-R)
low enough, you should use a low speed to write your data-CDs or DVDs. In the
past, the writing speed for best BLER values were indicated on the blank media,
but today this is seldom the case.
The BLER value is a statistical error value of digital media. The Red Book specification stipulates the highest acceptable value.
For comparative listening, you can create a second file, dithered to 16 bits in the
last position of the master insert of the virtual DAW mixer.
If you are working with a digital mixer, it is a good idea to create the greatest
possible bit depth – in this case 24 bits – when creating the mix master. Because
the S/P-DIF and AES/EBU digital transfer formats are limited to 24-bit integer bit
depth, it is impossible to process files on a 32-bit basis with external equipment.
The 16-bit DAT recorders are to be avoided nowadays; the DAT format is no
longer suited as a mastering medium. In addition, very few project studios have
high-resolution tape machines. For mixing down using an external digital mixer,
it is best to copy back over to the DAW. Here it is important to make sure that the
synchronization (wordclock/houseclock) settings are correct.
For good sound in the digital studio using a houseclock, the rule is: converters
clock themselves internally, are clock masters, and drive the clock distributor.
(If you own several different converters, then the most important converter – that
the mic and line inputs are fed through – should be internally clocked.)
If they are in the chain, devices like the TC Finalizer should be set to bypass or
used only as “technical” limiters for eliminating occasional peaks. This does not
contradict the high quality of the Finalizer, but avoids unnecessary compression,
which can be difficult to correct in the mastering process.
Be careful! When you are using S/P-DIF with your digital mixer, make sure that
the processing chain supports 24-bit word depth. Some audio interfaces – especially low-cost models – are limited to 16 bits. Such devices should be replaced
with 24-bit audio interfaces if they are to be used for mixdowns. WaveLab provides a bit-depth metering function that shows the actual allowed bit depth.
If you are working with an analog mixer, you can make a ½-inch master parallel
to the digital master. These formats are still accepted in some mastering studios
in North America. In Europe large mastering studios also support such formats.
Beware of the country-specific measuring standards of analog machines (USA:
IEC/Germany: NAB) and be gentle with tape saturation. With a digital mix master, high-quality 24-bit converters should be used. These also serve as clock masters for the DAW being used for recording. 24-bit or 32-bit floating point files
should be delivered on CD-ROMs. If reliable digital metering is available, I recommend keeping pop and radio music at an average loudness of not more than
-14dB/RMS during loud passages. This ensures that the mastering engineer’s job
can remain enjoyable. Over-compressed masters are very difficult to work on and
are difficult to shape. When mixdowns are louder than -14dB/RMS, it is difficult
to correct mixing mistakes.
Here is a summary of the most important points for delivering a mix master to a
mastering studio:
u Highest possible bit depth; Cubase & Nuendo: 32-bit floating point files; otherwise use 24-bit files.
u In situations where the computer uses internal 32-bit floating point processing, but you must go to 24-bit files for exporting, use either a brickwall limiter
without dithering or a simple limiter and dither down to 24 bits.
u Do not dither 32-bit files.
u Use CD-ROMs, DVD-ROMs (slow writing speeds), hard disks, or memory
sticks as transport media.
u For CD productions, use a sampling rate of 44.1 kHz. If conversion is necessary, use a high-quality sample rate converter (SRC) with internal oversampling. If this is not an option, leave the conversion to the mastering studio.
Sample rate conversion is a complex and CPU-intensive process, which can
only be done with oversampling (multiplication of the sampling rate) in order
to prevent rounding errors.
u Before mastering, avoid fades at all costs unless they are musically arranged
into the song. If fades are run through dynamic processing during mastering,
pumping and digital artifacts can result.
u Do not cut beginnings and endings. Many mastering plug-ins require short
lead-ins so that the “predict” function can work. Without the lead-ins, artifacts might be created during the processing phase. When marking areas to be
bounced or exported in the arrangement window of your DAW, it is smart to
leave small lead-in times instead of selecting the area start point exactly at the
beginning of the song.
u When using analog mixers, a lead-in can be used as a fingerprint for possible
de-noising during mastering.
u Label track files according to their numerical order on the album. (For example, 01-32Bit my song.wav.)
u Give the mastering studio a processing wishlist along with reference tracks, if
Please see Chapter 2: Using Metering for Monitoring the Three Dimensions.
The mastering engineer brings various mixdowns together into a unified sound
aesthetic. These mixdowns were often made weeks, months, or years apart and
were maybe even mixed in different studios by different engineers. The mastering
engineer carefully increases loudness and shapes the material according to the
Fletcher-Munson curve and makes adjustments so that the tracks sound equally
good on a car radio, in the kitchen, and on a living room stereo.
A personal bias can arise if one is too involved in the mixing process and should
be avoided. When mastering your own production, you should allow some time
to pass between the two phases. If possible, avoid mastering in the same listening
room that you mixed in; after all, one of the roles of the mastering studio is to
find potential mistakes that result in faulty acoustics in the mixing studio! If the
mixing studio has certain room modes and you master a track in the same room,
then the mistake will remain undetected.
You will find more information on the subject of mastering in my book “Audio
Mastering with PC Workstations.”
O GET THE BEST RESULTS with software-based mixing, there are a
number of technical issues to keep in mind. This applies to both the selection and use of equipment. First of all, I would like to look into the
advantages and disadvantages of analog summing. Recently, my readers have often asked me whether or not analog summing is useful or even necessary. This
is why I have decided to take up this issue here in the form of a thorough test.
The audiophile results are available on the accompanying DVD (in the folder
“Summing Unit Test”).
The Story of Analog Summing Units
Why does this equipment exist, and why is it supposed to be so good, when software mixing has become so inexpensive and convenient?
Over the past decade, purely digital working methods have also made their way
into professional circles – above all thanks to the ease-of-use and effectiveness of
working with total recall. At the same time, there are more calls for alternatives
to software-based digital summing. Critics claim that software-based mixes often sound flatter, less homogenous, and digital. Even before the reasons for this
were empirically examined, mixing consoles from Mackie – or, budget permitting – Neve, SSL, and others found their way back into studios for the purpose
of analog summing. The external mixing consoles are mostly used with neutral
settings (faders at 0 dB), in order to take full advantage of the total recall features of the audio workstation. The idea is to reduce analog summing down to
the essentials: high quality preamps that are summed with fixed levels. The three
test candidates are from Tube-Tech, Tegeler Audiomanufaktur, and Dangerous
Music and can be put into this “puristic” category. The concept can be summed
up with “16 inputs/ 2 outputs” (exception: TSM with 32 inputs). Some of the
devices (Dangerous 2-Bus, audient sumo, SPL MixDream and Neve 8816) have
parallel latency-free direct outputs for hooking up analog monitor controllers.
The parallel outputs for the aux or 2-track sends on the monitor controller makes
it possible to check the sound against the DAW mix outputs, to ensure that there
are no overs or sound discolorations as a result of reconversion. The analogsummed signal, which comes back into the computer via A/D converter can also
be heard at the DAW mix outputs.
The analog summed result should produce better depth and separation, sound
warmer and more analog, and should also help make the mix more cohesive.
In this context, it is interesting to note that – as far as I can tell – there is still not
much information on potential sonic disadvantages of digital summing available, with nebulous myths built up around the idea. It hasn’t even been firmly
established that a problem exists at all.
Prerequisites for Digital Summing
To understand the usefulness of analog summing, we should look into the potential problems of digital summing. The technical prerequisites for digital summing
with good spatial depth is the constant use of 32-bit floating point. Most CPUs
(AMD, Intel) – as well as most host applications (for example Cubase, Nuendo,
Logic, ProTools) -- operate with 32 bits internally (Motorola = 24-bit fixed point).
Using 16-bit or 24-bit files for offline processing such for normalization, fades,
effects, etc., as well as bouncing or exporting files with a bit depth of
Truncation under 32 bits, leads to loss of “finer” spatial information. In my experiinvolves ence, the 32-bit floating point data format helps to create good spatial representation. Although this is not yet scientifically proven, it has
removing been confirmed by many readers’ comments and is explored in depth
in the following section “More Bits for More Sound.”
inferior bits. It
is the opposite
of dithering.
In ProTools, the plug-ins operate with a 24-bit integer resolution,
whereas ProTools calculates with 32-bit resolution internally. This
means that each processing step contributes to a slight loss in depth
(to reduce the effect of this phenomenon, ProTools and Logic provide
a dithering option at the corresponding processing steps). If truncation occurs,
seldom does the loss in quality play a big role. But if truncation occurs too often,
the cumulated effect will be an increased flattening of the sound through the loss
of quiet (ambient) information.
For more information, see: “What is Truncation”.
If 32-bit floating point resolution is not used throughout, using analog summing
can help a flat mix. The more tracks that are digitally summed, the stronger the
effect of cumulative truncation.
Be careful! The depth of a recording is originally determined by the ambience or
reverb part of the signal (which is often recorded by extra ambience microphones
in large studios, in order to have more control). If this is missing at the recording stage or is not satisfactorily present, then low-quality reverb devices (which
cannot produce depth) will not help at the end. Perhaps the best way to illustrate
this phenomenon is to compare it with a cheap film production, where all dialogs
are recorded in dry rooms; they remain dry and unprocessed all the way through
post-production mixing. Regardless of whether or not the actors speak in a forest, a toilet, a factory, or out in the open, the voice is always completely dry. The
audience therefore has a permanent feeling of incongruity and flatness.
See Chapter 6 on “Dimension 3” for more information on this topic.
If all of the conditions for good digital summing are met, the analog summing
unit can still provide that little extra. Just as a Lexicon 960 sounds better using the
internal A/D-D/A converters than with direct digital inputs and outputs, good
D/A converters, and high-quality analog input and output transformers also have
their sonic charm that go beyond the necessity of analog summing. At this point,
sound design aspects come into play and go beyond the role of summing itself.
To really be able to judge the difference, the signal path must include a 100% neutral D/A converter, as well as summing amps that are just as neutral.
And now for the real advantages of analog summing.
Working with Summing Units
Looking into the past, we can see that analog summing has a clear advantage: You
reach your objectives more quickly. Because of the subjective nature of working
habits, it would be difficult to compare the results from 10 different engineers
working under time constraints on the same song under different conditions.
This makes it all the more important to point out this difference, since it is not
part of the test. Somehow, you obtain a good, homogenous basic sound faster
when you use analog summing, regardless of whether you are using a simple
summing amp or an analog console. My feeling is that on the digital level, you
have to work harder to create the same result. Although I realize that this viewpoint is controversial, I have had it confirmed many times through my own experience and by many other sound engineers. On the other hand, you cannot
expect to automatically obtain a good mix just because you are using analog summing. For users with good skills in software-based mixing, analog summing can
add that extra bit of spice to their mixes.
device varies significantly. Please refer to the details in the device descriptions.
Another advantage of analog summing is the possibility of taking advantage of
transistor or tube saturation and the resulting uneven harmonics, which add a bit
of “warmth” to the end result.
The Test
This effect does not occur during software-based mixdowns, as long as it is not
simulated with devices such as the “PSP VintageWarmer” or Steinberg’s “Magneto”
in group tracks. These units also offer strong saturation behavior when the incoming signal is increased. To do justice to the saturation aspect, a few examples
in the test are provided with high levels.
Note that there is another area of application that goes beyond the scope of our
test. This kind of equipment serves very well as a front-end for external keyboards.
For users who would like to stop using their consoles, but cannot because of their
arsenal of keyboards and expanders, high quality analog summing devices are a
good alternative with which to improve sound.
How can a kind of device be fairly compared, when there are such big differences
in terms of functions? How important is the use of analog summing for achieving
truly professional results? What the difference in working methods? As an advocate of software-based working methods, I have decided to clarify the mystical
all-rounder. Here I have tried to find the common denominator between the six
Another potential advantage is connecting external analog outboard equipment
in the sum or even in individual audio channels. With summing units without
master inserts, high-quality EQs, and compressors can be connected directly to
the outputs of the unit. Here, unnecessary critical reconversion (A/D) is avoided,
as opposed to inserting into the sum of the software mixer. This can make a difference when a large number of external gems are inserted into the internal mix.
Only the SPL MixDream provides a type of bridgehead into the analog world.
Each stereo channel has a switchable insert path. Accordingly, the MixDream
can be used with 8 stereo devices or even device chains in the 8 busses. If you
use this feature, you are saving 16 A/D conversions, which you would otherwise
have accepted with analog devices placed directly on the outputs and inputs of
the converters. This cannot be fairly compared, since the SPL device is the only
candidate with these features. For this reason, we will not take this aspect into
consideration. The MixDream is thoroughly well designed, with direct outputs
and a “no mix” button for every stereo channel; when activated, this reroutes
the signal of the inserted outboard device back into the DAW when you want to
insert the device just once in the mix. This saves you from having to crawl behind
your studio rack in order to re-cable.
Another potentially positive aspect are extra functions in the master section such as
compressor/limiter, stereo expander, or M/S matrix of some devices. In this area, each
The six candidates
block diagram internal mixdown
Block Diagram
Internal Mixdown
Stereogroup Drums
A determining factor for the choice of equipment is worldwide availability and
filling the role of the common denominator: “16 in – 2 out.”
Reverb 5
Reverb 4
Reverb 3
Reverb 2
Aux 1
Aux 2
Aux 3
Aux 4
Aux 5
Reverb 1
Guitar solo
Stereogroup Guitar
Stereogroup Keyb
Lead Vox 2
Lead Vox 1
Stereogroup Vox
Overhead L
Overhead R
Stereogroup Overhead
Tom 3
Tom 2
Tom 1
Stereogroup Toms
The Test Candidates
Rim open
Snare (solo)
Rim gated
Monogroup Snare
test candidates, and have developed a test involving an unadulterated and fair
comparison where your judgment is also taken into consideration. The partly
very subtle variations in sound, having the most varying attributes can quickly
wear out our powers of imagination. With the help of the platform-independent OMF file, you should form your own opinion as to whether or not analog
summing is necessary! Get an idea for yourself with loudness-corrected A/B
comparison between a software-based mix and analog summed mixes. But first
read on.
The equipment: Neve 8816, SPL MixDream, TSM from Tegeler Audiomanufaktur,
Dangerous 2-Bus, Audient Sumo, and the Danish SSA 2B from Tube-Tech. I
would like to thank the manufacturers and distributors for quickly and easily
providing these devices!
!! No Inserts !! No Send FX !! Fader 0dB !!
Insert: only technical limiting for over protection
Setting Up the Test
The group outputs and the effect returns of the test song are routed to the outputs
of the RME audio interface, which in turn go directly into the inputs of the summing units. The cable between the RME converters and the summing unit inputs
Out 3 > Summing Mixer I/P 3
Out 1 > Summing Mixer I/P 1
Out 7/8 > Summing Mixer I/P 7/8
Stereogroup Toms
Out 5/6 > Summing Mixer I/P 5/6
Reverb 5
Reverb 4
Reverb 3
Out 15/16 > Summing Mixer I/P 15/16
Reverb 2
Aux 1
Aux 2
Aux 3
Aux 4
Aux 5
Reverb 1
Guitar solo
Stereogroup Guitar
Out 13/14 > Summing Mixer I/P 13/14
Stereogroup Keys
Out 11/12 > Summing Mixer I/P 11/12
Stereogroup Vox
Out 9/10 > Summing Mixer I/P 9/10
Stereogroup Overhead
Lead Vox 2
Lead Vox 1
Overhead R
Overhead L
Tom 3
Tom 2
Tom 1
Snare (solo)
Rim open
Rim gated
Monogroup Snare
In Nuendo I like to use group channels for insert effects such as EQs, compressors, and stereo expanders (which I will explain later in detail). In order to
internally and externally sum the same signal, an internal mix is created with
a drum group without insert or send effects. It is set at 0 dB. This way, the subgroups “Snare,” “Toms,” and “Overheads” – as well as bass drum and bass – can
be separated from the drum group in order to be able to route them to individual outputs without changing the volume relation in the mix. Using a group
compressor or an aux-send for reverb in the drum group, or a level other than
0 dB would falsify the results between internal and external summing because
of level differences.
block diagramAnalog
analog summing
Block Diagram
DAW Out 2 > Summing Mixer I/P 2
The test host was Nuendo 3 on a dual core PC running at 3.4 GHz with 2GB of
RAM connected to two 8-Channel RME ADI-8 QS converters. The excellent
quality of these converters ensures that no disturbing coloration will occur on
the conversion level. Good converters are a basic prerequisite for using analog
summing units.
Analog Summing Device
DAW Input 1&2 via RME ADI-8 QS
The same levels are used in the summing of both mixdown variations in the block diagrams.
has gold-plated contacts and are made by Mogami. For the returns, the stereo
outputs of the summing units are routed into the stereo inputs of the RME converter with VoVox cables. For reasons of sound neutrality, I have avoided using
the Universal Audio 2-610 for line inputs.
The same channel configuration was used for all units:
Input 1:
bass drum
Input 2:
Input 3:
Input 4:
Input 5&6:
Input 7&8:
Input 9&10:
Input 11&12: keyboards
Input 13&14: guitar
Input 15&16: effect returns
I then created two different mixdowns with each summing unit in order to
fairly show the different possibilities of the devices. When the devices were not
equipped with their own level control, I used the RME audio interface mixer
“Total Mix” in a few of the summings to increase the levels of all channels from 0
to 6 dB, in order to test the sound saturation behavior of the units when provided
with a stronger signal. These mixes are indicated with “+6 IP”.
All results have – at least theoretically – identical level and panorama behavior.
Only potential level tolerances of the summing units could have any affect on the
mix behavior.
Here you can see the device settings, the basis for the twelve different summing
results (six devices, each with two mixes):
Neve 8816
Mix 1: Channel level: 3’o’clock, Master: 1 line under full
Mix 2: Channel level: Right position; 1 line under full
Mix 1: Limiter: off; St-exp.: off; Transformer: on, M:0 dB
Mix 2: Limiter: on (18); St-exp.: 3; Transformer: on;
Tegeler TSM1
Mix 1: Output: 0 dB
Mix 2: +6 IP; Output: -6 dB
Dangerous 2-Bus
Mix 1: + 6 dB-switch off; Output: +6 dB
Mix 2: + 6 dB-switch on; Output: 0 dB
Audient Sumo
Mix 1: Mix Master: 0 dB, Master: 0 dB; no sum. comp.
Mix 2: MixMaster: 0 dB, Master: 0 dB; with sum. comp.
Tube-Tech SSA 2B
Mix 1: Output: 0 dB
Mix 2: +6 IP; Output: -6 dB
Total Mix with +6 dB emphasis starting in channel 9 up to
channel 24 which are connected to the analog summing unit
Finally, in order to create fair conditions for an A/B comparison, all the files were
brought to the same loudness level without any dynamic-altering intervention.
The order of the A/B comparisons in the test file (OMF) is purposely not the
same as the order presented here, to prevent bias. Also, sometimes the softwarebased mixdown is on the upper channel and sometimes it is on the lower one
of the OMF project. In order to neutralize your ears, please listen entirely to the
short “distraction sequence” provided between each A/B comparison. This serves
as a way of “resetting” your ears for the next comparison. Also, it is best to set the
track display in your DAW so that you cannot see the actual waveforms – this
way you can really concentrate 100% on what you are hearing. Now, if you follow
the rules, your results will be very meaningful! Do not jump back and forth between the examples. You can do this when choosing your favorite after you have
filled out column 12 in the evaluation list.
You will find your personal evaluation list as a PDF file on the DVD in the folder
“Summing Unit Test.”
The Test Procedure
After connecting the first summing unit and with high expectations, switch on
the input list button of the Nuendo input track, this should deliver the enhanced
master output of the summing unit. If you hear cracking or popping sounds: is it
jitter? Sync-error?
After quickly checking the sync settings, you will see the solution: laFact or fiction: tency crackling! The mix project, which can easily internally deal with
of 256 samples (6 ms), must be increased to 512 samples
Analog a(12latency
ms) for a crackle-free mixdown. Just to be sure, after restarting
summing units Nuendo, I repeated the latency test. The internal mixdown requires
37% CPU load (Nuendo indication) at 6 ms for crackle free operation,
save CPU. compared with 42% CPU with the same latency in external mixdown
mode. With 12 ms, the external mixdown requires only 30% CPU load and is
completely crackle-free.
Fact or fiction: Analog summing units save CPU
As far as Nuendo – and therefore also Cubase – this often heard argument does
not hold up. It can be assumed that other hosts react in the same way. Routing
to individual outputs requires more computing power and therefore higher latency.
Level adjustment is fortunately very easy with all the devices: set everything to 0
dB and at the end the right level arrives in Nuendo. Extremely simple!
Another requirement for this test is the possibility of switching at least the first 4
channels to mono, so that snare, bass drum, and bass will also arrive in the center.
All manufacturers have understood this need and have answered it in various
The Dangerous 2-Bus provides mono switches for all 8 stereo busses, SPL’s
MixDream for the first three channel pairs (i.e. for inputs 1 to 6), Audient Sumo
and Tube-Tech have mono switches for the first 2 stereo pairs, the Neve 8816
has panorama pots on all 16 inputs, and the Tegeler TSM provides for automatic
mono switching whenever a mono input occupies only the odd-numbered input
jack of a stereo pair (i.e. channel 1, 3, 5, etc.). This can be a bit inconvenient if
you want to permanently cable the device and be able to use different configurations.
All manufacturers should be praised for adhering to level and pin assignment
standards. All sub-D connectors are assigned according to the Tascam standard,
so that error-free cabling was possible right away.
Device Particularities
Tegeler Audiomanufaktur TSM1
The TSM has a chic vintage design and is a simple and basic version of an analog summing amplifier with 16 stereo inputs (via 32 balanced mono jacks) and
tube-refined summing output. The only particularity, made apparent by the test,
was that mono-cabled channels appeared louder in the mix than in Nuendo. In
Nuendo I use the -3 dB panning law, which lowers mono signals in the center
position by 3 dB (see chapter on panning). This is the most neutral way of working and is also used by high-end analog consoles. It is assumed that in the TSM1,
a different panning law is used. On a practical level it is simply annoying, when
you are trying to get the same results internally and externally. If the TSM is normally integrated in a setup, then the respective channels would simply be mixed
3 dB quieter. The 3 dB level steps of the damping output control works quite accurately. Two ECC88 tubes are used in the TSM1.
Tube-Tech SSA 2B
The Danish Tube-Tech SSA is very similar to the Tegeler summing device. The
16 balanced XLR inputs are summed together. The 2 U rack height is a purely
physical dimension; the powerhouse with 8 tubes (two ECC82s, two ECC 83s,
and four ECC 88s) should have a least 1 U headroom clearance to allow for heat.
My impression is that the tubes not only warm up the control room, but also the
mix. The gold-plated incremental output controller has a range of +/- 10 dB in
steps of 0.5 dB (0 to 3 dB), 1 dB (3 to 6 dB), and 2 dB (6 to 10 dB). Inputs 1-4 can
be set to mono with a toggle switch.
Dangerous 2-Bus
The American 2-Bus summing device is tubeless and is a bit more lavishly
equipped. For every stereo pair there is a +6 dB gain switch as well as a mono
switch. Next to the mix output is a parallel monitor output for connecting the
summing device directly to the monitor matrix. In addition, an input for cascading with another 2-Bus is included. The external power supply gives a nice, solid
impression with a robust 7-pole plug to connect with the device. Output level is
adjusted incrementally in 0.5 dB steps with a potentiometer with a range of – 4
dB to + 6 dB. The device’s levels are also very precise.
Audient Sumo
The Sumo summing amplifier by Audient is conceptually a little different from
the others. It is just 1 U high, which is made possible with a mini fan, and is fortunately so quiet that it poses no problem when mounted in a rack. Sumo also allows for switching the first 4 channels to mono. “Mix Gain” allows you to reduce
gain before the compressor and limiter. Unfortunately, the pot is not incremental.
The Sumo is the first test candidate with an insert in the master section, outfitted
with an on/off switch. It can also be switched pre or post dynamic section. These
features were not included in the test, since they could not be compared with
the others. The single-band compressor boasts all typical compressor parameters:
threshold, ratio, attack, release, and make-up gain. Next in the signal chain is a
peak limiter, which is controlled by a threshold potentiometer. If I now put myself
into the shoes of a mastering engineer, I realize that personally, I use only technical limiting in the sum channels and deal with dynamic processing in a separate
stage of the work. Ideally, a specialist should carry this out. If you do not share my
opinion, then you can always use the compressor for soft “pre-compression.” In
one of the examples, that is exactly what I have done. The output level “Mix” can
be adjusted from minus infinity to +6 dB with the Mixmaster potentiometer. I
miss a reliable incremental pot here. While working, I noticed a left-right imbalance of around 0.2 dB. A nice feature is the separate monitor output, complete
with mono switch, gain, and switchable source (EXT IN). Finally, an A/D converter with a sample rate of 192 kHz is available as an option. The converter was
not used in the testing.
SPL MixDream
The particularity of the MixDream is the very clever “bridgehead” concept; in
addition to the simple 16 in - 2 out principle, the MixDream also provides direct
out channel inserts and can be connected to up to 8 analog stereo devices. An
insert switch in each stereo channel turns them on or off or – if you choose to use
an external device individually – can be taken out of the summing. This position
is labeled “No Mix.” As opposed to direct connection to external devices through
D/A-A/D converters when using a summing unit, this prevents sound-degrading reconversion since conversion only occurs once with the summing signal.
The “Mix” and “Monitor” outputs are XLR connectors, all other connectors are
Channels 1 to 6 can be switched to mono with toggles. The MixDream is expandable and with three units, you can even use it in a surround setup. The “Expansion”
switch routes the additional units into the summing signal before the inserts.
The master insert is switchable and can be adjusted from -20 dB to +7 dB. The
summed signal can be treated with a stereo expander, but for me, this would be
part of a mastering process. (Very detailed results can be obtained in M/S processing when controlling stereo expansion with multi-band compression.) The
last element in the chain is a switchable limiter for keeping peaks under control.
The Lundahl transformers at the output – which are also switchable – add typical
warmth and silkiness to the sound.
Neve 8816
Neve’s summing unit has the most structure among the devices that were tested
– it is actually a real mixer with a mastering/monitoring section, which can also
be used as an analog summing unit. A complete description of all possible uses
is beyond our scope here, so we will just list the features briefly: a 16 in / 2 out
mixer with panning and level potentiometers as well as cue and cut/solo switches
for every channel. The cue function provides an independent headphone mix for
overdub sessions. “Cut” allows you to turn off individual tracks. You can create a
solo group by switching the “Cut/Solo” switch in the Master Section to solo. The
device remembers the “Cut” and “Solo” settings of the individual channels, allowing you to create complex A/B comparisons between complete different channel
combinations. When used uniquely as a summing device, the potentiometers
should not be moved. With the correct input and output levels, the channel potentiometers can be left at the position all the way to the right, without clipping
in the mix. This is certainly the fastest total recall method. Caution: In some cases
with high-level input signals this can lead to a slight analog compression in the
stereo sum of the summing unit. The panorama law is fixed to a 3 dB attenuation at the center position by the manufacturer. Because of this the mix relationships are not changed when a mono instrument is placed in the center. Without
this attenuation, the mono signal, which is equally distributed between L and R,
would result in a 3 dB gain increase when placed in the center compared to side
placement. A particularity of the 8816 is its automation capability, inspired by
big mixing consoles. This is controlled with a USB-connected computer. In the
classic total recall fashion, the potentiometers can be set to the values displayed
on the computer monitor. These virtual pots then disappear as soon as the values
between the physical pots and the real pots match. Most users probably do not
have the patience to deal with this when they are only dealing with summing. The
panning and gain automation can be very helpful if the 8816 is used as a frontend for a collection of synthesizers. The ability to solve phasing problems by stereo reduction of widely spread layer sounds, or the ability to place any keyboard
more in the center is a very useful feature for this use.
To summarize the functions of the master and monitor section: headphone jack
with potentiometer and switchable summing signal; speaker toggle (A/B); talkback level; talkback mic; master insert; insert mix return; insert can be switched
to M/S thanks to the internal M/S encoding and decoding circuitry; 2 track return as well as iPod jack on the front panel; high-quality A/D converter with DSD
outputs and sampling rates of up to 192 kHz available as an option. With the
additional fader pack the volume potentiometers can be used for aux-send level
control. The 8816 is also expandable.
Test Results
Please only look at this list after you have done your own listening tests, so that
you are not biased. Please print out the “Evaluation List.pdf ” before starting with
the comparision test.
Here is the sequence arrangement in the OMF file:
Ch. A
Tube-T.-1 int
Ch. B
2-Bus 1 int
Sumo-2 int
8816-1 int
2-Bus 2 SPL-1
Tegeler-2 Tube-T.2 int
Sumo-1 Tegeler1
Without going into the details concerning the differences of each summing unit,
there is a small acoustic advantage with analog summing. Characterizations such
as “unity” or “the mix feels tight” come to mind. The question is whether this
improvement comes from analog summing itself or from the sound-coloring nature of the analog circuitry. In my experience, proper mastering allows for much
greater sound optimization in terms of these sound characterizations.
It is up to you to determine if the investment in 16 converters, good cable, and a
summing unit really makes sense for you. The previously mentioned aspects concerning working faster and avoiding unnecessary A/D conversions using analog
outboard devices is also to be taken into consideration.
In conclusion, using analog summing devices can enrich the results of a mix both
as a result of the individual character of the device, and through avoiding cumulative truncation. A lack of room ambience during recording cannot be compensated with analog summing. Consistent and continuous use of 32-bit floating point files along with good reverb devices make professional software-based
mixes possible.
To examine this topic – which is essential for both production and mastering
– we will take a brief look into the basics of digital techniques.
PCM – The Principles of Digital Audio
PCM stands for Pulse Code Modulation and is the most common way that audio is digitalized. The .wav and .aif formats are based on PCM. PCM technique
involves cutting an analog signal into slices, called samples. The audio CD format
uses 44,100 samples for every second of audio material, while audio for DVD
video employs a sample rate of 48 kHz. Higher sampling rates are used for formats like DVD-Audio. The highest reproducible audio frequency is half of the
sampling rate. This means that with a sampling rate of 44.1 kHz, audio information of up to about 22 kHz can be recorded. Every sample consists of a 32-bit
subframe, while usually at least 16 bits are used for representing amplitude.
Bit Resolution
The audio CD standard calls for a so-called 16-bit word length. Each of these
audio bits can be either a 1 or a 0. Therefore there are 216 possibilities. 216 equals
65536 steps (2x2x2x2x2x2x2x2x2x2x2x2x2x2x2x2), half of which are discrete
positive amplitude values and the other half negative values. A series of these
values can be represented as ascending and descending steps.
More bits for The reason audio frequencies can only be reproduced at up to
the sampling rate is that in order to be represented, a fremore sound half
quency (= oscillations per second) needs both a positive and
– what you a negative value. At 44,100 samples per second that results in
an upper frequency limit of 22,050 Hz.
should know
about bit depth
and truncation
To make the bit pattern clearly visible, visually enlarge a .wav
file in your DAW until the individual steps are recognizable.
With 24-bit resolution there are 16,777,216 steps (224) – increased bit depth creates a logarithmic improvement in dynamics and precision.
Comparing 16-bit and 24-bit resolution, there is 256 times as much information
for a relatively small increase in data storage!
By multiplying the bit depth (also called “word length”) by 6, you get the maximum dynamic range that can be represented. Using this formula, 16 bits gives a
range of 96 dB and 24 bits increases the dynamic range to 144 dB.
With 32-bit floating point processing, 24 bits plus 8 so-called exponents are used.
A detailed description is beyond the scope of this book.
“Stairs” in a digitalized file in WaveLab
For the purposes of determining which bit resolution to use, here are a few basic
characteristics of 32-bit floating point processing:
u As demonstrated with the calculations above, 32-bit resolution provides a great
deal more processing precision and dynamic range than 24-bit resolution.
u As long as audio data remains in the 32-bit floating point format, it is almost
impossible to produce overs. Even signals that exceed 0 dB can be precisely
processed and stored in the exponents of the 32-bit floating point data. With
24-bit integer data (fixed point), levels above 0 dB are represented as 0 dB
chains, leading to artifacts.
u If you work entirely with 32-bit floating point resolution, you do not have to
worry about potential overs. You only have to avoid signals above 0 dB at the
very end of the processing chain, when bit-reduction is necessary to convert
the data to the target format (24, 20, or 16 bits).
u Most of the well-known PC and Mac-based DAWs (Logic, Cubase, Nuendo,
WaveLab, DP, ProTools, etc.) as well as nearly all good VST and DirectX plugins work in the 32-bit floating point resolution.
u All current Intel and AMD CPUs also calculate with 32-bit floating point.
Because of the uneven number of bytes with 24-bit files, it actually takes less
processing power to operate on the 32-bit basis. By not using 32-bit floating
point, you are often actually using more computing power. With mixing projects having more than 80 tracks, it is simply the increased data flow caused
by hard disk read and write operations – along with PCI bus data capacity
– which limits performance.
u As long as the amplification remains ‘normal’, that is, the sum of the signals
does not exceed 0 dB, the 32-bit float format behaves like a 24-bit integer calculation.
That adds up to many advantages. But in order to understand the fundamental
acoustic advantage we need to understand truncation and the consequences of
cumulative truncation.
What is Truncation?
Truncation is the removal of lower value bits through the reduction of word
length. When we have a 32-bit calculation result, which can only be saved in the
24-bit format, then the mantissa is simply deleted. Opinions differ on whether or
not the sound actually suffers during this process. In many situations, it would
be difficult to clearly distinguish between a dithered and a truncated file in an
A/B comparison if there is a decrease from 32 to 24 bits and this decrease only
takes place in a single generation. But problems begin to arise if you repeatedly
decrease the word length in a computer-based mixdown. Try to create a deep,
transparent mix from 40 tracks with a program that does not consistently operate
on the 32-bit level. It is simply not possible because the many bit-depth decreases
remove increasingly more subtleties necessary for such a quality mix. When you
have mixed together the first 8 tracks, it still sounds good, but the quality will
deteriorate with each further track. This is the consequence of cumulative truncation.
Why is High Bit Depth So Important for Sound?
I have already provided some explanation on the importance of high bit depth
for sound quality in the previous section on truncation. When CDs first came
out, music production was still somewhat of an elite sovereign territory – at least
nobody could produce a CD at home. Production generally took place on high-
quality analog equipment before the material was digitized at the very end of the
editing chain and processed onto the CD master. If you carry out this last step
very carefully – using input dithering and paying attention to every detail – you
can get excellent results with 16 bits. However, back then, 20 different digital calculations were not made along the way. Nowadays, it is frequently only the microphone, the pre-amplifier, and the monitoring path that are analog. Everything
else happens exclusively in the digital domain and requires dozens of calculations.
Almost every digital calculation entails additional decimal places that should be
represented as precisely as possible. For example, a simple 1-decibel level change
can require an extra 7 decimal places, making the 32-bit format absolutely necessary for accurate processing. The problem with short word lengths is therefore
the inaccuracy inherent in numerous successive processing. Just think of every
little change in a mixing project which requires calculation: fades, cross fades,
level changes, sends, EQing, insert effects, and many more.
Who cares about the seventh place after the decimal point?
Our hearing has the ability to unconsciously perceive the subtlest sounds that
carry information relating to the three-dimensionality of a recording. When
3D information components occur at minus 95 dB, we cannot consciously hear
them. However, we clearly miss them when they are no longer present. If you
imagine that the lower value bits, which keep being discarded due to the usage of
low bit depth, always carry these subtle intricacies it is no wonder that the recording sounds flat.
What Happens When a 16-bit File is Saved as a 32-bit
The 16 empty bits will simply be tacked on to the bit word. At first, these empty
bits have no acoustic impact. I only do this when I receive 16-bit material for a
mix, so that I do not need to give any more thought to bit depth if any further
editing and offline processing is necessary. Once I save my work after the first
processing (for example: normalizing, changing levels on individual clips, fades,
offline processing, etc.) a “real” 32-bit file has been created that uses the full bit
depth when necessary.
Why Record in 32-bit Format When the A/D
Converter Only Supports 24-bit Resolution?
This is the same situation as with 16-bit files. Eight “empty” bits are simply tacked
on. The advantage for the sound quality first becomes clear when the first processing is undertaken and when you want to save the file after offline processing.
Even if you are using the 24-bit format, the internal calculations are still being
done in 32 bits, but are saved and therefore truncated to 24 bits. Normalization is
certainly one of the most commonly used offline processes. When you normalize
a 24-bit file, the calculations are done in 32 bits and the least significant bits are
therefore thrown out since the 24-bit format you are using cannot use this higher
resolution. And – as previously pointed out – today’s computers actually work
more efficiently with 32 bits, therefore giving you the further advantage of saving
CPU resources.
The technical foundation for obtaining satisfying results requires correctly
dealing with bit depth and understanding concepts such as dithering and truncation.
Summary of a Few Basic Rules Relating to Bit
u Always use the best possible resolution: 32-bit floating point wave files or even
better 64-bit if already available in your DAW!
u If you need to periodically save between processing steps (offline processing
such as level changes, fades, normalization, DC offset removal, and any other
processing) in order to compensate for less computer power, you should convert the original source material to 32 bits wherever possible.
u Always dither when converting to lower bit depths.
u Generally, dither as seldom as possible.
u Only use 32-bit-floating point plug-ins. This can be checked with the WaveLab
bit meter.
u Don't truncate (shorten word length) without dithering.
u Deliver high-resolution material to the mastering studio.
pen a 32-bit audio file in WaveLab and insert the plugin you want to test into the WaveLab Master Rack. The
plug-in has to be working, of course. In other words, the
settings should not be in a neutral position. If all of the 24
bits light up (green), as well as the blue “Internal” light, then
the plug-in is operating on the 32-bit floating point level.
If this is not so, you should check to see if there is a similar
or better 32-bit plug-in which will do the same job. Using a
16-bit file, the Bit meter lets you easily see that only 16 bits
are being used, as long as the file is not being processed in
any way. Even if you just barely change the level, all 32 bits
will be used.
The WaveLab Bit meter lets you check the bit depth
being used. If you play a 16-bit file without any
processing, the meter will show a depth of 16 bits.
As soon as you change the level – even by a very
small amount – the meter will show a 32-bit depth.
When Can We Leave the 32-bit Domain?
u If you want to connect an external device via AES/EBU, you should dither the
internal 32-bit floating point signal to 24 bits, before routing it to the outputs.
u If you want to burn a Red Book test CD as part of a music production, you can
export it for this purpose by dithering it to 16-bit resolution. If you are satisfied
with the result, export the master for further processing in 32-bit format.
u If you want to use a plug-in which you know only works in 16 or 24 bits, find
the corresponding resolution through dithering. I would prefer not using such
Tip for Mastering
In the last stage of mastering, aim for the target bit depth by dithering. No more
changes may be made to the master thereafter as these will immediately occur in
PC-based 32-bit resolution and the optimal usage of 16 bits through dithering
will be ruined. In addition, you must re-dither or you will truncate to 16 bits as
the internal calculation takes place at 32 bits.
In this book, the emphasis is on practical approaches. For more information on
dithering and other basic concepts, please refer to the book “Audio Mastering
with PC Workstations.”
Nowadays dithering refers to adding very quiet noise to the signal in order to
modulate the least significant bits combined with bit mapping. This increases
the representation precision. Both techniques are used in the process of shortening the word length of files. The main purpose of dithering is to ensure a good
representation of the sound when an audio file is converted to a lower bit depth.
Simultaneous bitmapping ensures the optimal use of the available bits. This is why
in mastering, once the last dithering has been carried out, the audio file should
not be manipulated in any way.
Contrary to sample rate reduction, where all the advantages of higher sampling
rates are lost when the rate is reduced, good dithering can maintain most of the
sound quality when bit depth is reduced.
Acoustics can be divided into building acoustics and room acoustics.
Building acoustics protects us from “unmusical” neighbors, traffic sounds, and
other nuisances from our surroundings. Room acoustics deals with what happens to sound inside your studio; it is about the acoustic characteristics of your
studio rooms. When I talk about studio acoustics here, I mean this latter aspect,
having to do with the interior acoustics of listening or recording studio rooms.
problem frequencies (room modes). A three-dimensional graphic representation
shows any selected mode in your room.
There is always a lot of discussion concerning studio acoustics. Although it is not
the aim of this book to go deeply into this subject, I would like to touch upon the
matter here so that you are aware of the importance of studio acoustics and how
it effects your work. Many engineers underestimate the significance of acoustics
with regards to their ability to hear and judge their own work.
Loop these tones and play them back moderately loudly (above 70 dB/SPL) and
move around your usual listening position within a radius of about one and a half
meters (about 5 ft).
Here’s a little experiment you can do; take a measuring stick or a tape measure
and measure your listening room (W x L x H). Assuming that most rooms are
quadrilateral, go to the following website:
and calculate the room modes of your listening room. Please make a note of the
With the Signal Generator in WaveLab you can generate sinus tones corresponding to modes in your
listening room
Open up WaveLab (demo version provided on DVD). With the Signal Generator
(Tools), create a sinus wave at each of the frequencies that you noted.
If your listening room has a resonance frequency at 82 Hz
(room mode) and the bass drum in the track you are working
on has this same characteristic core frequency, the following
will occur: in your listening position, the bass drum might
sound much too loud, but if you move one meter (3´3´´)
away, the same sound will be much too quiet; or the other
to help calculate
between frequency and
wavelength, try the
following website:
way around. How can you possibly judge your work under such circumstances?
If the acoustics are not optimal, then you should know the defects as well as possible. For example, you can use a bass-sweep to record the hills and valleys in your
listening position. A simple SPL meter is fully adequate for measuring. On the accompanying DVD you will find a level-calibrated bass sound that goes through
the bass range in semitones. The sound consists of the combination of a short
tone and a long tone in order to make it possible to get a feel for each frequency’s
decay time. Print out the pdf “Frequency Chart” and use it as the basis for your
notes. Set a standard SPL meter to “C” weighting and make sure it is in “slow”
mode. Hold it vertically in your listening position and play the bass sounds with
a uniform loudness of at least 78 dB/SPL. Make a note of any deviations.
The results could look something like this:
37 Hz
+5 dB
45 Hz
-2 dB
55 Hz
and so on.
If an A in a bass line appears too loud, you can glance at the list and see right away
that your room (in the listening position) makes that particular tone twice as
loud as others. This way, you can learn to work with any acoustic weak spots you
might have in your listening room. If you have any other listening positions in
your room, for example for guest listeners like a sofa, then I recommend carrying
out a separate set of measurements for this listening position as well.
Typical SPL meter that
can be purchased for
around $50
The widely used foam insulation – produced in a near-infinite number of varieties – is effective against slap echoes and overly long decay times in the middle
and high frequency ranges. However, if overused, the result can be an irregular or
dead sound in these frequencies. The above-mentioned bass and mid-range frequencies are little affected by this material. But that is exactly where many problems arise while recording and mixing.
How to Live With Imperfect Acoustics:
u If you cannot invest much in the acoustic construction of your studio, the next
best thing is to be perfectly familiar with the acoustics of your studio. Listen
to many reference recordings to have a clear, confident feeling for what your
speakers and studio acoustics sound like.
u Also use reference headphones – just like the acoustics of your studio, you
should be as familiar as possible to how they sound.
u A good graphic analyzer is not only good for finding problems in the bass
frequencies – it is also simply a valuable tool for obtaining good results. In addition, when used properly it will train your ears.
You will find further information concerning metering in the section “Using
Metering for Monitoring the Dimensions” on page 64.
Even though the cost of investing in a modern music production facility has gone
down thanks to computer-based systems, there are a few things to consider in
order to make good mixes. Even with a great deal of knowledge and experience,
it is very rare to make a good mix with just a laptop and a pair of headphones.
Choosing good studio equipment will definitely bring you much closer to your
Choosing Speakers
The same rules for studio acoustics apply to choosing speakers: the “truer” the
signal that reaches our ears – and in this case, over the entire signal path (D/A
converters, cable, mixer or monitoring matrix, cable, amps, cable, speakers) – the
better the conditions for judging and acting on the sound.
This can be compared to a sieve; the coarser the sieve, the grainier the result and
the more difficult it is to shape it.
In your listening room, speakers act like a box in a box. The small (loudspeaker)
box interacts with the bigger (room) box. That makes the choice difficult. The
typical demo room in a studio equipment department
rarely has good acoustics. Testing speakers in your own The finer the resolution
room also can adulterate the results, depending on of the chain of
speaker position and especially on the room’s characreproduction, the finer
the possibilities for
judging and acting on
the sound.
The important characteristics for choosing speakers are:
Spatialization (Dimension 1 = Creating a stereo field)
(Dimension 2 = Frequencies)
Spatial depth (Dimension 3 = Depth)
These factors depend on such things as speaker box design, crossover network
circuit design, and much more. Another important criterion is good ‘’listenability’’ making it possible to listen without rapidly becoming fatigued. At the same
time, a studio speaker should never have the pleasant-sounding and well-rounded characteristics of a hi-fi speaker.
If you do not work in a listening room which has been professionally measured, I
recommend using products which have already proven themselves in professional environments. If good speakers sound bad in your room, the reason is probably the quality of your room and not that of the speakers. In this case, change the
environment, not the speakers.
The Basics of Setting up Speakers:
u Set up the speakers within a 60° angle (30° left of the middle and 30° right of
the middle).
u Set up the speakers symmetrically in the room.
u Keep enough distance between the speakers and the walls (at least 50 cm
(20´´), refer to the set-up instructions of your speakers).
u Ensure that the listening distance is sufficient (with near-field monitors, at least
1 meter (3´3´´)).
u The tweeters of both speakers should be the same height as your ears when
sitting in your monitor chair (ca. 120 cm (4´)).
u Avoid unnecessary objects (such as CRT monitors) in the sound field between
the speakers and the listening position.
u Use spikes under near-field monitors to insulate against structure-borne sound
– this will tighten up the bass sounds.
u Whenever possible, use large, heavy speaker stands which do not resonate
with the speakers.
u Avoid “loudness” circuits (as is common in hi-fi amps) in the monitor signal
u EQs are the last resort for correcting when all else fails. Only the best components should be used. It is better to make precise corrections with parametric EQs than with graphic equalizers, which have a greater tendency to create
phase errors and muddy sound.
One of the most frequent speaker setup mistakes I have seen is the use of “practical” workstation tables with side flaps for speaker placement. In most cases, the
speakers are much too close to the ears and too far apart.
The same rules apply to both your room and your speakers: in the end, the most
important factor is your familiarity with your listening environment. I once met
a producer – now successful internationally – who achieved excellent results in
the basement studio he had at the time. He mixed with plastic active speakers
that cost less than $100 (used nowadays as multimedia speakers for computers). Combined with headphones, he could perfectly compensate for the lacking
sound quality of the speakers. Of course, this is not an ideal situation, but it was
certainly very expedient.
By front-end, we mean all the devices that are placed along the signal path before
the actual recording process. These principal elements are microphones, mic preamps, A/D converters and cables. But it also includes the dynamic processing of
compressors and EQs. Here we will only address the principal devices.
Unfortunately, a bad choice of front-end devices often is a major source of problems. The quality of every single component – including cables – is so important that I recommend investing in ONE top-quality signal chain rather than
several mediocre. For example, if you can invest $2,000 in a mic preamp, then
choose one that only lets you adjust quiet/loud – but does this perfectly – rather
than a machine with numerous extras like a compressor, an EQ, DeEsser, which
are good, but not excellent. All of the fineness, quality, depth, and resolution of
a good recording chain will make every further processing step in your DAW
easier. A finely-recorded lead vocal recording is obviously much easier to place
in the mix than one badly recorded with a $150 dollar microphone, a third-class
preamp, and the computer’s built-in 16-bit converter.
You (or your clients, if you have a mix gig) determine the level of quality of the
entire production with the front-end chain! A top-quality computer-mixed production can only be achieved with the best-recorded signal quality possible.
Cable is often neglected in discussions about the front-end. Recently, I had a revelation when recording vocals with completely new cable bought from VOVOX.
The front-end – already excellent quality – got a clear boost with improved depth
resolution and finer, cleaner high frequency transmission. Please do not cut
corners when cabling your front-end chain – good cable is a worthwhile investment.
Since we are focusing on a mixing procedure that should also be able to make
a lower quality recording sound good, we should also look into the back-end,
which is also important for the mixing process.
Just as the name implies, the back-end describes the signal chain from the medium (computer hard disk) through the D/A converters, the monitor matrix or
mixing console, the power amp, the speakers, and of course the cable paths.
D/A Converters
The D/A converters are the first step along the back-end signal chain and play a
significant role in terms of representing depth and space. Built-in converters are
unacceptable for any ambitious mixing project. When choosing a high-end audio interface, do not shy away from comparisons, even though it may not be easy
or practical to install and remove different devices in your computer. The 24-bit,
96 kHz audio interfaces are the absolute standard nowadays. Because there are
only a few different converter chips on the market, the differences in quality lie
mainly in the analog circuitry of the converters and the quality of the clock.
If you plan to purchase a new interface, you should first create a needs-profile,
outlining the number and format of inputs and outputs necessary; then see what
is on the market that fulfills these criteria. Your budget will further narrow the
choices so that you will only have a few devices to choose from in the end. In any
case, look for the appropriate products in specialized music shops, not computer
Minimum requirements for PC or Mac audio interfaces:
u ASIO driver for Windows systems, Core Audio driver for Mac. ASIO stands for
“Audio Signal Input/Output” and is an open standard developed by Steinberg
that allows for low latency operation.
u Reliable operation down to 3 ms latency. (Low latencies are only important in
the recording and production phase. To free up CPU resources, you can set
latency as high as 46 ms while mixing. With higher values, you will feel a slight
delay when working with your DAW.)
u Audio interface drivers are crucial for reliable operation. You can usually see
on the support page of the manufacturer’s Internet site if updates are posted
regularly. Forums often can quickly give you a good idea of both quality and of
customer satisfaction.
u 24-bit S/P-DIF connections. Cheap interfaces often support only 16 bits. Be
sure to check this!
u Wordclock input and output, in order to be able to synchronize other digital
devices. Audio interfaces without wordclock can only be used alone, without
other devices.
What is latency?
atency, or input latency, refers to the
amount of time the computer takes to
record and reproduce signals with a given
audio interface. Low latencies are necessary for
sending the recording input signal to monitors
or headphones, without giving the musician a
feeling of delay. Sound needs 3 ms to travel one
meter (3´3´´) through a room. A bass player who
is 3 meters (10´) from the drummer – or the side
fill monitors providing the drum sound – has to
deal with a latency of 9 ms. Depending on the
musician’s level of training, latencies starting at
6 ms will be perceived as unpleasant.
If your DAW’s CPU meter is over 50% because
you are using many tracks or have plug-ins that
pull a lot of CPU, then low latency levels can
result in crackling in the monitor path. In this
case, either remove the plug-ins or increase the
latency in order to ensure flawless operation.
Cubase and Nuendo offer the function Constrain
Delay Compensation, this allows you to simply
switch off plug-ins, which in turn increase the
latency. This is helpful when you want to lay
down another track in a song where the mix
is nearly finished. Some DAWs like Cubase and
Nuendo add the so-called output latency with
the input latency. Output latency is only present
when CPU-intensive plug-ins are in the master
bus, group channels, or VSTi channels. The
output latency is brought into play in order to
apply automatic latency compensation over
the entire mixing console. The Constrain Delay
Compensation button only applies to the CPUintensive plug-ins in the above-mentioned
channels, since the processing time (for automatic
latency compensation) cannot be calculated in
advance, as is the case with audio channels.
The cable you use is as important in determining sound quality as the speakers.
It is well worth it to purchase high quality cable. Bad cable can downgrade the
best devices to mediocrity. To limit costs, concentrate on the monitor signal path.
With active monitors, this consists of only two pairs of cables from the converters to the monitor controller and from the monitor controller to the speakers. In
addition, I recommend having high quality instrument and microphone cables
as well as good preamp-to-converter cables on hand. I have included some useful
Internet links to a few manufacturers in the index.
Monitor Controller, Monitor Matrix,
and Mixing Console
Most often, mixing consoles serve as switching and adjustment units for speakers, conveniently bringing together features like talkback, machine controller,
and multi-channel recording in one relatively inexpensive package. A modern
computer-based studio can also function without a mixing console and is often
even better without one.
When you imagine how little technical cost and effort goes into a single pre-amp
or monitor controller circuit in a 24-in-8 inline mixing desk costing little over
$1000, then doing without this equipment is comprehensible. It is obvious that a
monitor controller with just a small number of choice components dedicated to
the few tasks that are done outside of the computer is better for the job.
Another significant point against using a large mixing console is the extremely
large, acoustically unfavorable surface, which can add a comb filter effect to the
speaker signal.
In addition, using two mixing consoles makes for a confusing workflow: what
should I manage with the external console and what should I manage with the
computer-internal DAW console? These thought processes become burdensome
while recording an entire band live. The existence of two
mixers is especially irritating for novice studio users.
Important Information!
Since the internal mixing console is a complete represen• Do not control monitor levels digitally within
tation of its analog counterpart, concentrating on ONE
the computer! Reducing level on the digital
console greatly simplifies concentration. For multi-track
domain reduces bit depth – the number of bits
recordings with bands, you just need to make sure there
used – and inevitably leads to a flatter sound.
are enough inputs and outputs, and to set up mic preamps
• Always send a full-level signal out of your
as well as headphone amps outside of the computer.
From a purist point of view, it is better to use a really
good-sounding, high quality monitor matrix or monitor
controller that has integrated talkback. The SPL Monitor
DAW into the converters and reduce volume to
the desired level only with high quality analog
• In monitor systems like the TC/Dynaudio Air
series or the JBL LSR4328 & 4326P, remote level
control is analog – after the D/A conversion – in
order to take advantage of full bit depth on the
digital level.
Talkback Controller (MTC) is one of the pioneer products in this field. Nowadays
there is a large choice of products on the market.
If you need several different headphone signal paths, then simply set this up internally in the computer and send the signals over separate outputs, going directly
into the headphone preamps. You can replace the preamps from an obsolete mixing console with high quality individual units. This way, you will surely achieve
better results than by using inexpensive mixers.
Remote controllers are a relatively new kind of device that arrived along with
the development of audio computers. They vary in price between about $50 and
$100,000. The basic idea is a replacement of mixing console panels, in order to get
away from the mouse-oriented operation of DAWs. It might also have a lot to do
with outward appeal, though. Beyond the comfort of using well-built encoders,
these devices have only two real advantages for me:
Power Amplifiers
• While working on projects where frame-by-frame synchronization is important, such as postproduction and film music, a jog wheel is indispensable.
As the last link in the chain before the speakers, the power amplifiers also influence sound quality significantly. Priority should be given to Class-A amplifiers,
which have a great deal of reserve power. A power amp should always be able
to deliver much more power than a speaker can take. This actually protects the
speakers rather than damages them. Thanks to the reserve power, no distortion
artifacts are created by low end artifacts and peaks. This distortion would cause
imprecision to the signal. And for the ears, the advantage is higher fidelity impulse reproduction.
Digital Monitors – the Purist’s Back-end
Now there is an exotic possibility to reduce the sound-affecting influences to a
minimum with digital speakers. The only professional versions that I know of
so far are Air series from TC and Dynaudio, and the JBL LSR4328 and 4326P.
These are active monitors with digital inputs (AES/EBU) and high-quality 24bit D/A converters, which convert the signal at this last possible stage to analog
for the speaker cabinet. Using digital signals this way ensures that sound-influencing aspects of analog equipment – including cables and monitor controllers – are no longer part of the back-end chain. Volume is handled by a small
remote control unit that is networked with analog circuitry in the speakers, so
that reducing volume never reduces the bit depth of the digital signal (which
would degrade sound quality). This way, the speakers are nearly noise-free and
are very faithful at reproducing the sound. Software allows you to calibrate the
monitors easily to your listening environment. In my studio I have a 5.1 setup
using the Air 20.
Remotes and Controllers
• When writing over automation, touch-sensitive faders (these can record over
previously written automation when touched and stop recording when released)
are extremely comfortable to use.
In my opinion, controllers make sense only when you have at least 32, if not 64
channels in front of you. Flipping through the “pages” of 8-channel controllers
quickly becomes tiring, so that in practice you end up using mostly the mouse to
control your DAW mixer. Many people who use small controllers become quickly disappointed and realize that they work faster with the mouse.
The economic advantages of large controllers through a small workflow acceleration is only worth the investment when the studio is being used continuously.
Controllers can be divided into two groups:
Pure controllers, which have neither physical inputs or outputs nor converters.
The second group are controllers that include internal DSP power, or one or
more features of standard digital mixers.
For budgetary reasons, purchasing a controller with several good mic and line
preamps and AD/DA converters can make sense. This is the case when such a
controller has high quality mic/line preamps and converters, which would be
costly as separate outboard equipment. If you use this controller to adjust your
monitor volume as well, then make sure that the volume adjustment is carried
out by good quality analog circuitry. Otherwise – as discussed in the section on
monitor matrixes – you will have the problem of lower bit depth. An example of a
helpful tool here is a high-quality volume controller such as the Volume 2 by SPL.
With the concept of dealing uniquely with high-quality analog volume control,
this device is a forerunner on the market.
My personal experience from observations of many project studios has led me
to conclude that the combination of a minimal controller and systematic use of
working methods using keyboard shortcuts is more effective than an inefficiently
used controller and the constant, time-consuming switch between mouse and
controller. In the high-end range, there are a few interesting solutions, such as
the Icon console from Digidesign, the Euphonix 5MC-EuCon or the Smart AV
console. For a professional-sounding mix, a good mouse is sufficient, but obviously a large mixing console or controller is more fun to use. In a professional
studio situation, where an engineer works 10 hours a day and 25 days per month,
the advantages in terms of ergonomics and protection of the eyes and the hand
(which holds the mouse) are clear. Detaching ourselves from the visually-laden
environment of the computer monitor helps to obtain more balance between the
senses, to the benefit of our sense of hearing. This can be a factor in the quality of
the production. In addition, the purchase of large controllers can be determined
by rather odd economic factors such as being able to justify studio fees. The show
must go on!
Using Metering for Monitoring the Dimensions
Graphically monitoring the audio signal with high-quality metering such as
graphic analyzers, spectrometers, goniometers, peak meters, and correlation
meters provides significant help in all stages of production. To use such equipment correctly, it is important to follow the rule “first listen, then look” to prevent
our ears from being overshadowed by our eyes in judging the sound. Placing the
equipment on the side, so that it is not constantly in our field of vision, is a good
way of achieving this. With the head faced forward in the center, we can get a
good aural impression of things, which can be more or less confirmed with a
glance to the metering on the side. This also makes a graphic analyzer a valuable
autodidactic tool. Regular cross-checking teaches us about important frequencies where various instruments have a positive or negative effect.
Since this book deals with software-based solutions, let’s concentrate on inexpensive – but not any less professional – computer-based metering solutions.
PAS metering
Along with the host-internal analysis tools that range from “so-so” to “ok,” there
are also some very powerful stand-alone solutions. The metering from PASProducts.com and the Pinguin Audiometer are leaders in this field. Mac users
can also check Spectre by audiofile-engineering.com. To illustrate the various
functions and the practical use of such tools in the studio, we will look at the
Pinguin metering. It is widely used in mastering, broadcast, and large recording
studios. Metering from PAS-Products has nearly the same range of functions for
a fraction of the cost and is a very viable alternative to the Pinguin. In comparison
to the Pinguin, only the stereo meter is missing from the PAS-Metering. All other
explanations apply to both products.
The Pinguin can either work alongside the DAW on the same computer (PC
only) or on an external PC. When installed on an external PC, the capacity of the
audio workstation computer is not affected. Sufficient for this would be an older
750 MHz PC with at least 64 MB RAM and a full duplex audio interface with a
Routing from the audio computer to the metering computer
Pinguin PG-AM 4.5 Audiometer:
Vertical left: Peak and Loudness measurement;
Middle lower area: Correlation meter;
Lower right: Goniometer;
Main part: Graphic or spectrum analyzer
16-bit or – even better – a 24-bit S/P-DIF input. This would be enough for using
the Pinguin or the PAS-Metering smoothly and without latency problems.
Since I use an RME Hammerfall audio interface in my computer which includes
a DSP mixer (“Total Mix”), I can route the signal for the monitor controller in
the summing buss digitally at 0 dB to the S/P-DIF output. This output is then
connected to the metering computer with a 75-Ohm S/P-DIF cable. In my studio
I also use an RME Multiface for the metering computer. That way, 6 parallel surround channels can be routed in via ADAT optical cable for Pinguin surround
metering. The RME Multiface can also be used very well with laptop computers
with the Cardbus interface using the PCMCIA slot, making it possible to use the
Pinguin as a mobile metering solution for jobs outside the studio or even for live
Notes on surround metering:
ou can see on the chart [preceding page] that I
have lowered the levels of surround channels 3
to 8 by 4.8 dB each and have routed them to the S/PDIF output. The reason for this is that for surround
productions, I like to hear the mix of the 6 surround
channels in the listening position while seeing the
results on a spectrum analyzer. That is why L & Ls is
routed to the left and R & Rs to the right, while C & LFE
are in the center of the panorama. The level reduction is
necessary because the sum of the 6 individual channels
in the stereo display should only exceed 0 dBFS in
exceptional cases.
It is important that the signal is digitally routed into the metering and that the
level is exactly the same as the master level which goes into the D/A converters
before the studio monitors. If your S/P-DIF output is already in use, you can use
a simple Y adapter, which is available for both coaxial and for optical S/P-DIF.
Peak and Loudness Measurement
What is the difference between peak and RMS levels, or peak and loudness? The
maximum level of a digital recording, measured in dB = decibels (peak) is represented by the greatest amplitude (wave size). For example, when the waveforms of
two concert guitars perfectly match at a particular point, then amplitudes resulting
from the addition of both individual amplitudes can be very strong without us perceiving this sound event as being twice as loud as the other parts of the song. The
peak value does not tell us much about perceived loudness.
The difference is therefore that loudness is measured in dB/RMS (root mean
square), which represents the density of a recording and therefore tells us something about its loudness. Looking at the waveform of a “loud” song is reminiscent
of a thick sausage. The thicker and more “sausage-like” the recording looks (with
the same waveform magnification), the louder it is.
Here are some practical uses of a peak meter:
u Checking the peak headroom of the sum throughout the entire production
process, providing information for limiter settings in the master bus.
u Measurement of the operation, reliability, and precision of new plug-ins. (Are
the level indications on the knobs correct? Is the limiter really keeping overs
from coming through?)
u Stereo level adjustment.
u Indication of level relationships between different elements; for example, between bass drum and snare. (When the bass drum’s amplitude is greater than
the snare’s, then it is probably because of an error in judging the levels or perhaps a problem with the bass frequency reproduction of the speaker.)
Most devices that measure the amplitude peaks also provide loudness measurement. Generally, the display shows a second bar which is colored differently or
otherwise differentiated and which runs parallel to the peak value display. For example, the WaveLab peak display is green on the outside and blue on the inside.
Loudness should indicate the “thickness” and therefore the perceived loudness
of the audio material. Loudness is always lower than the peaks. The important
factor in loudness measurement is the readability, which corresponds strongly to
the ballistic of the display as well as the ability to objectively be able to read and
interpret the information displayed.
Loudness measurement can be used:
u To check the average loudness of tracks. (Is the mix too compressed?)
u To check the loudness consistency of different takes of a track.
u To check energy distribution between L & R in stereo.
u To prevent mixing and mastering too loud.
There is a lot less to say about goniometers. In Pinguin it is called stereo meter,
which is self-explanatory. Goniometers provide an intuitive way of visualizing the
amount of stereo in the sound. If you send a mono signal into the left channel, it
will show a line at a 45-degree angle on the left side; a line in the middle indicates
a mono signal in the middle and a mono signal sent into the right channel produces a line at a 45-degree angle on the right side. If you send a dry mono signal
into the center (for example, the side click of the snare) along with a long stereo
reverb, you will see a line in the middle followed by a rounded shape. It is best
to examine this within your DAW. The goniometer is used to quickly and intuitively get an idea of the stereo width in different mixes, which should be unified
and consistent. Goniometers also provide information about the differing stereo
widths of keyboard sounds and all stereo signals in general. In addition, phase
problems are clearly represented. If the signal doesn‘t look like an elongated egg
along the vertical axis, but instead is horizontally oriented, then there is a phasing
problem. The Pinguin meter also provides a numerical readout that indicates the
left-right weighting in degrees.
Spectrum Analyzers
Spectrum analyzers are certainly the most important tools for visualizing sound.
A good spectrum graph reliably shows the frequency distribution of a recording. The most common frequency resolutions are one band per octave and three
bands per octave. The frequency spectrum is split up – like a graphic EQ – and
shown in separate bands. In my opinion, reliable and intuitive work with such
resolutions is not possible. The spectrum analyzer in the Pinguin PG-AM-Pro
displays in half-step resolution (12 bands per octave) and offers a number of
different settings and display possibilities. The PAS software can even be set to
quarter-tone steps (24 bands per octave). A particularly intuitive mode is the
“logarithmic pink noise flat” setting, which displays pink noise on a logarithmic
frequency resolution as a flat horizontal line. In this mode, the frequency distribution of recordings are particularly easy to interpret, since a good-sounding
recording has a nearly flat frequency spectrum with roll-offs on the low and high
The Pinguin Goniometer shows Stereo Width.
ends of the entire frequency range. From a scientific analysis point of view, the
“logarithmic pink noise” setting is questionable, since it does not provide any
precise measurement information. But what we need in this case is an intuitive
and readable display that gives us a sure way to recognize if specific frequency
regions are over or under emphasized – this is why this display mode is very useful. A good graphic analyzer uncovers problems in the infra-bass frequencies that
would escape even the best subwoofer.
The quick and intuitive graphic representation of information is particularly helpful, making it possible to easily recognize parts of the signal that are part of the
music’s rhythm or not. Working with analyzers taught me to put lowcut filters on
all tracks to rid them of unnecessary bass artifacts and thereby create more space
for the bass. With a glance, an analyzer will show you if a mix was properly done.
The peak hold function allows you to gather very precise frequency information
by using the mouse to highlight specific areas, making accurate use of EQs possible. If you listen to an instrument group spot solo, it is quickly evident if the lower
mid-range is overemphasized with respect to the lead vocals, making it possible
for you to effectively remedy the situation. On the other hand, you can also see
where frequency ranges are not being used, so that you can use these for events
that had been previously buried. All of this teaches you about frequency/pitch
relationships and the sweet spots of each instrument. When used live, the loudest
frequency is automatically displayed, which is very useful if there is feedback and
you need to know which frequencies need to be attenuated. Good metering is
highly valuable for the motivated sound engineer.
Correlation Meter
Sine wave, fully correlated (0 degrees)
Sine wave, right channel shifted by 90 degrees;
Meter in the middle
Correlation meters show the mono compatibility of a stereo signal. In other
words, the correlation between the two channels of a stereo signal is measured.
The display is always split between a green area that is scaled from 0 to 90 degrees
and a red area, scaled from 90 to 180 degrees showing the phase situation. If the
left and right channels add up, we stay in the green area. If the phase of the signal
is shifted, then the sum of the signals produces a smaller value and is then shown
in the red area.
In the three illustrations on the facing page there is a stereo sine wave that is
displaced by 0, 90, and 180 degrees, along with the corresponding correlation
Sine wave, right channel shifted by 180 degrees;
red area
Correlation meters used to be a must in every studio and were usually either
mounted on the console meter bridge or simply built in. Because mono players are less common than they once were, mono compatibility is mostly only
an issue in television production. Even when the target medium is a vinyl LP,
you should still pay attention to correlation! Since LPs are mono in the bass
frequencies, recordings with phase cancellation cannot be used for records
because needles cannot follow such movement. While working in the studio,
glancing at the correlation meter should always be accompanied by switching
the monitors to mono. Only when your listen to your stereo mix “collapsed” to
mono can you really hear if potential phase displacement (in mono) truly results in acoustically relevant phase cancellation. If the mix falls apart when you
switch to mono, then maybe you have panned too much low frequency sound
or you have too much phase cancellation. Using the mono switch is referred
to as physical phase cancellation, since both channels are mixed together on a
physical level. Acoustic phase cancellation – in other words, the cancellation
that you hear in a room – is always weaker because the signal in the room is
naturally de-correlated.
For more information on the potential causes of phase cancellation, please see the
section “Dimension 1: Horizontal = Panorama’’ on page 107
Pinguin Spectrometer
Pinguin Spectrometer
For people with an affinity towards all things nautical, this device might remind them of a “fish finder.” Once you get used to it, it is a good alternative to a
spectrum analyzer, with the last 20 seconds of audio being spectrally analyzed.
If the playback is stopped, the spectrum analyzer also stops, so that you have
time to check over problem areas with the computer mouse. In the line at the
lower edge of the window, information on frequency and level in 0.2 dB steps
is provided. The horizontal axis is the time axis from 0 to 20 seconds and the
vertical axis represent frequencies from 14.57 Hz to 22.354 kHz. Levels are color
scaled. Yellowish-red lines along the horizontal axis could for example indicate
continuous overemphasis in a particular frequency band, inciting the engineer
to find a solution to the problem. Vertical lines indicate short sound events. The
spectrometer is an effective tool for helping you see particularities in the audio
material along the time axis, making it possible for you to immediately see level
inconsistencies and ensure a good frequency distribution.
Extra DSP Power for Software-based Mixing
If your goal is to mix complex productions with many tracks on your computer
without having to constantly resort to rending your files as offline processes, then
you will definitely need extra DSP (digital signal processing) power. I see today’s
music computer as a tape machine, MIDI sequencer, and fully automated mixing console. The sound production stage is not a part of the mix process since
the MIDI files have already all been converted to audio. For all intensive signal
processing that goes from a rudimentary lowcut filter on a single track to the
use of various send effects, using extra DSP power is unavoidable. A native (i.e.
computer-internal) plug-in is by nature a type of compromise between its performance and the CPU power needed for this performance. It is relatively easy
to program an excellent native compressor, which is so performance-hungry that
you can only use a few on your computer. On the other hand, it is difficult to create a good-sounding compressor that doesn’t bring your computer to a crunching stop when you use 25 instances of it. To compare with standard studios, I
consider extra DSP as outboard equipment. Regardless of which host application
you use – all midrange consoles costing around $100,000 have one thing in common: they all have EQs and dynamic effects for basic work. But even here, the
most important tracks are processed with the best outboard equipment.
Hard Choices
The following section provides an overview of four established, cross-platform
DSP systems: the UAD-1 by Universal Audio; PowerCore by TC; the Waves
APA; and the SSL-Duende. To be complete, I should mention Liquid Mix from
Focusrite. But I have simply not had the chance to thoroughly test this device and
develop my opinion about it. If you are interested, you can surely find information
and test reports on Liquid Mix in the Internet and in specialized magazines.
UAD-1 PCI Card from Universal Audio
UAD-1 Card
The UAD-1 card is a rather small card for PCI or PCIe slots. It relies on a very
powerful DSP processor that was originally designed for video rendering. The
card operates completely with 32-bit floating point resolution and therefore permits seamless integration into most host environments. Only in-house Universal
Audio plug-ins work on this card, which is available in several different variations. The only fundamental difference between the different UAD-1 card products is the software that is delivered along with the hardware. The UAD-1 varies
in price between around $500 and $2000 and can be expandable to up to four
UAD-1 cards in one computer. Unlocking and updating additional plug-ins is
easy. The plug-ins that you own on one card will automatically be available on all
further cards ($500 version).
The latency which results from sending the signal to the card and then bringing
it back to the computer‘s CPU are equivalent to the system latency x 2 and are automatically compensated in Cubase, Nuendo, and most other hosts. When automatic latency compensation is not available, then a plug-in tool (delay compensation) can help out. The more power required of the card, the more it also requires
in terms of CPU resource management power. The main focus of the product
is the emulation of legendary classic equipment such as those from Universal
Audio, including, for example, the 1176-LN compressor and the LA-2A Tube
Limiter. These are still little treasures that can be found in well-equipped analog studios. Recently, the plug-in choice was broadened with a series of modeled
Neve legends (1073-EQ and 1081-EQ, as well as the Neve 33609 Compressor).
I will go into further details in the chapter regarding plug-ins (page 203).
The UAD-Xpander system is a UAD-1 card built into a small housing for laptops
with the new ExpressCard 34 connection (successor to the PCMCIA cardbus
slot). With a PCIe adapter, you can also connect the UAD-Xpander to a desktop
PowerCore FireWire by TC
Included in this category are the mastering tools, such as the MD3 compressor,
taken from the TC System 6000 (with a price in the five-figure range), or the
reverb algorithms of the System 6000 VSS3-Reverb. PowerCore is aligned more
with the high-end professional market with technically modern tools.
UAD Xpander with ExpressCard 34 connection
Shortly after the release of the UAD-1, TC brought out the PowerCore PCI card,
followed by the PCI-Express version, PowerCore Compact, PowerCore FireWire,
and PowerCore X8. The PowerCore system is also expandable and is based on
150 MHz Motorola chips that operate in the 24-bit integer format. The Compact
version has two Motorola DSPs, the PCI-Cards and the FireWire versions four
DSPs and the X8 – as its name implies – is based on 8 DSPs. Depending on the
plug-in, the resolution goes up to 48-bit integer. The 24-bit integer format is the
only negative point; the PowerCore does not operate on the basis of 32-bit floating point format like the UAD-1 card does.
See page 79: “Incompatibility between UAD-1 and PowerCore?”
Many very high-quality plug-ins from third-party manufacturers such as the
Oxford series from Sonnox are available. The base price begins at about $500$800 for the PCI and Compact versions. The flagship X8 costs around $1500.
All versions are delivered with a basic assortment of plug-ins. You should plan
to budget more for the more interesting plug-ins, which are optional. You will be
rewarded with top-professional tools, which will turn your computer into a serious audio workstation.
The PowerCore series behaves similarly to the UAD-1 DSP unit when it comes
to latency compensation. Plug-ins can be unlocked with the serial number of the
PowerCore being used and is valid for one device. Automatic unlocking occurs
only when two or more PowerCores are connected to a single computer, in which
all the licensed plug-ins are unlocked for all of the PowerCore DSP units.
The FireWire PowerCore is happiest when it is the only device connected to your
FireWire port. Combining the SSL-Duende or a FireWire audio interface should
be tested individually. One way of avoiding conflict would be to add additional
PCI FireWire cards for additional FireWire devices, as long as slots are still free.
Incompatibility Between UAD-1 and PowerCore?
Both cards generally get along absolutely perfectly together.
Nevertheless, from a practical point of view, a few things should be taken into
consideration, which stem from the architecture of the cards. The vintage plugins from the UAD-1, which operate on a 32-bit floating point basis, sometimes
do not provide a way of controlling output levels (for example, the Pultec EQ)
or output ceilings in order to reliably control levels and peaks such as what you
would usually find in modern, digital devices. In itself, this is no problem, since
the 32-bit floating point resolution can easily deal with overs well above 0 dBFS.
But if you put some emphasis on the UAD-1 card’s Pultec EQ and then route
an “overdriven” 32-bit signal into a TC PowerCore plug-in, then you may have
problems. PowerCore’s inputs can only handle 24-bit signals, even when they
are internally calculated at 48-bit integer, as is the case with the MD3 software.
This is why you should always make sure that PowerCore’s inputs are not being
overdriven (check the input over display). I recommend that you work with some
headroom or use a reliable limiter with activated 24-bit dithering in front of the
PowerCore along the signal path. You can also use a CPU-saving leveler in front
of the PowerCore plug-in and reduce the level enough that the PowerCore plug-in
no longer shows any input overs. (Try out the FreeG plug-in by Sonalksis included
on the DVD in the software folder.) The input level adjustment of the PowerCore
plug-ins should not be used for level reduction in this case, since the level reduction occurs on the 24-bit level and cannot cleanly process the 32-bit overs.
I do not wish to stir up dithering hysteria at this point. If at one point, truncation
is carried out (because of a lack of dithering), the mix won’t immediately fall flat.
Truncation only becomes a problem when it is carried out repeatedly. There is
always a little room for interpretation in the debate concerning truncation vs.
too-frequent dithering. In the end, your ears should decide!
The PowerCore should be considered as an external device. If you route an external digital effect device into your setup, you are also dealing with just the 24-bits
provided by the AES/EBU standard.
With the APA DSP series, the well-known plug-in creator Waves has created an
innovative DSP system. APA stands for audio processing accelerator. There are
two models available: The APA32 is a 19” 1RU device which – strangely enough –
has 30% less power than the ½ rack sized APA44-M. The latter costs about twice
as much as the APA32 and retails for around $1,500; two plug-ins are included.
The ingenious networking concept allows you to not only cascade multiple units
to one computer, but to also share one or more APAs between multiple computer
workstations, giving good reason to create fights between workstation users for
DSP power! When using more than one APA, an Ethernet switch is necessary.
Calculation occurs in 32-bit floating resolution, meaning that this format can be
continuously used within most DAW environments. Only the additional Waves
plug-ins are processed by the APAs. At present, all Waves plug-ins are not yet
compatible with the APA (Waveshell capable). An APA44-M is capable of handling 26 Renaissance channels or 6 IR1 Reverb units. Installation is easy, as long
as a 1000 Mbps Ethernet port is available. Latency can be set in the control field
and is at least twice as high as the system latency you are using. Higher latency
provides more reserve power and better dropout protection, but make for more
sluggish handling with values of over 2048 samples (46 ms).
SSL Duende
The legendary British high-end mixing console manufacturer Solid State Logic
recently discovered the computer market and has brought out the Duende 32-
APA44-M by Waves
Duende 19“ DSP 1RU by SSL
Firewire and
System Requirements
he limited data throughput of the FireWire
format requires a separate FireWire bus per
FireWire device. If you plan on using more than
one Duende or a Duende along with a PowerCore
FireWire or a FireWire audio interface, you
should plan on a different FireWire bus for each
device with separate PCI or PCIe FireWire cards.
Otherwise, you will be faced with a high degree
of system instability.
bit floating point DSP unit. The Duende is not a DSP
unit that emulates the SSL character; it is the real SSL
The Duende allows you to process 32 mono channels
in SSL quality at 44.1 kHz. Along with the legendary
SSL stereo bus compressor of the 4000 series, the DSP
platform also includes channel strips and extras from
the E, G, and C series. Like the TC PowerCore, the
Duende is connected to the computer via FireWire.
It is cross-platform (Mac & PC) and the basic version
costs about $1500, including a solid array of plugins. SSL is progressively creating a pool of optionally
available plug-ins.
I will go into more detail concerning a few SSL plug-ins in the plug-in chapter.
Offline Processing as an Alternative
to DSP-supported Plug-Ins
By offline processing, we mean processing intermediate results, which are not
real-time. It is called “offline” processing because it does not occur in real time,
as would an insert plug-in for example. The various DAWs have their own words
for this function (bouncing, export audio, or rendering), which actually can be
thought of as a mixdown with automatic re-import of the processed file. This can
be laborious and time-consuming. Offline processing is easier, since you can do
the processing right from the main window of your DAW where the file to be
processed is located. Nuendo and Cubase provide a special function in that one
or more audio files in the project can be simultaneously processed. This can be a
track, an individual event, or a complete block of tracks. These offline processes
are saved in each calculated event and can be undone – even out of order – with
the “undo” function.
Offline processing in Nuendo and Cubase is called up via right-clicking
(when an audio event is selected) and choosing Plug-in or Effect. Here
is an example of the offline processing a lowcut EQ
This way of working is not as ideal as when using all plug-ins in real time, but
does provide a reasonable and very good alternative to external DSP units or
limited CPU power. For example, you can assign your favorite EQs to a keyboard
shortcut and then process a complete 16 track vocal group offline. Other DAWs
provide similarly structured functions for offline calculation.
Most readers are familiar with typical mixing console functions. Since modern
DAWs now offer very complex routing possibilities, I would like to go into the
basics of mixing consoles so that “mixing console novices” will be able to understand the terminology we are using.
The world of studio mixing consoles can be divided into two categories: in-line
and split-consoles. Both mixer concepts are designed to record up to 48 tracks si-
to “line input” on both kinds of mixing consoles so that they can be used as
full-fledged audio channels for the mix. Most consoles provide some kind of
possibility for switching between recording and mix modes. The mix mode
routes the tape returns to the input channels and frees the tape return channels,
so that they can be used for additional inputs, keyboards, or effect returns. A
24 in 24 console can manage 24 channels during recording and 48 for mixing,
since a separate control room mix – independent from the input channels and
recording levels – is not necessary during the mixing process.
In-line Consoles
The Nuendo/Cubase – Process List handles event-related offline processing history.
multaneously with good sound and appropriate levels, and to channel them into
a recording medium as well as providing independent control room and headphone mixes. In most project studios, people don’t usually record more than one
track at a time. This is why many self-taught audio engineers have a difficult time
understanding the entire concept on which complex mixing consoles are based.
In addition, the simultaneous co-existence of several mixing consoles can further
add to the complexity (the computer-internal software-based mixer, an analog
or digital hardware mixing console, and often an audio interface mixer, such as
the RME Total Mix). If there are no clear work procedures, then even seasoned
professionals can become filled with doubt and confusion. This is why I believe
that two good microphone preamps and a good volume control are worth more
than a mediocre mixer, and advocate getting rid of unnecessary hardware mixing
consoles in favor of a clearer, more logical, and more effective setup. Nowadays,
a good DAW can handle complex live studio recordings with 4 or more separate
headphone mixes and all other necessary functions.
As far as I know, there is no counterpart to the in-line mixer in the virtual world,
so I will simply point out the more obvious differences. Tape return channels are
optically and physically integrated into the input channels on an in-line console.
The most apparent characteristic of an in-line console is the fact that there are
two channel faders and two pan pots per input channel. Generally the main fader
is slightly larger. During recording, it allows you to adjust the level of the recorded
signal and in mix mode, it controls the tape return signal. In recording mode,
the in-line channel handles the control room mix; in mix mode, it can be used as
an extra channel for FX return, or additional inputs, for example. Most consoles
are designed so that the input or main channels are better equipped than inline channels. For example, a main channel might have a four-band parametric
equalizer and eight aux-sends. Most in-line consoles are very well designed and
offer many “flip” possibilities, in that functions can be assigned in various ways.
If you are working with such a console, you should study the user manual until
you have understood every function and you have the block diagram in your
head. That will improve your ability to spontaneously react to all situations that
an engineer might encounter in the course of a working day.
Each console is split into a channel area for the input channels and an area for
tape return channels. This is necessary for adjusting the sound and levels to be
recorded with the input section, as well as providing an independent control
room mix based on the tape return. The headphone mix is always captured
pre-fader from the tape return signal path, so that the musicians can hear what
the band has already recorded when they need to overdub. Tapping the headphone signal pre-fader makes it possible for the engineer to change the control
mix without affecting the headphone mix. The input channels can be switched
Split Consoles
Split consoles are somewhat easier to understand, since all channels are nicely
laid out next to each other and there are fewer flip functions that tax our understanding. In recording mode, a classic split console uses the input channels to adjust the signal being recorded and the tape return channels – generally located on
the right side of the console – are used for control room and headphone mixes.
Like in-line consoles, the configuration can be switched to mix mode, which
routes the tape returns to the main input channels and frees up the split section
– previously used as tape return – for other purposes.
Since the surface representations of the virtual mixers in Nuendo and Cubase
are almost identical to hardware consoles, it makes sense to use Nuendo as an
example to take a closer look at split console design.
The only thing that is missing on the virtual level is the switching between recording mode and mix mode. The reason for this is that switching is no longer
necessary – unlike their hardware counterparts, software mixers can have practically an unlimited number of channels available. The input channel section can
just be clicked out of sight with the mouse. That would be somewhat wasteful for
a hardware mixer.
The Cubase and Nuendo Split Console
The mixer is split into the following kinds of channels:
u Input channels
u Audio channels
u ReWire channels
u MIDI channels
u VST instrument channels
u Effect channels
u Group channels
u Output or master bus
u Audition bus
Before we can start working, the application must know what audio hardware we
will be using and how we want to use the available inputs and outputs. With F4
you can open a dialog box concerning VST connections, input and output configurations, as well as connections to outboard equipment. The number of tracks
you can record to is the same as the number of inputs on your audio interface. If
you are using two stereo monitors and want to have two stereo headphone paths,
then the outputs of an eight-channel audio interface might be configured like
Stereo monitor 1
Stereo monitor 2
Headphone Amp 1
Headphone Amp 2
Device-Port 1 & 2
Device-Port 3 & 4
Device-Port 5 & 6
Device-Port 7 & 8
The inputs might be configured like this:
Device-Port 1
Device-Port 2
Device-Port 3
Device-Port 4
Device-Ports 5 & 6
Device-Ports 7 & 8
SPL Channel-One 1
SPL Channel-One 2
SPL Gainstation 1
SPL Gainstation 2
Line In guitar preamp
Line In turntable
The Input Channel Section now shows the number of channels you have configured. These input channels are statically configured. This means that each input channel corresponds to a single physical input. Beyond general system configuration, there are no additional assignments necessary. The purpose of input
channels is simply to control the level and sound of the input signals that are
being recorded. This is why these channels can be compared to the recording
channels of a split console in recording mode. We cannot hear anything because
the fader determines the recording level and not the control room mix level. One
advantage of this concept is the option of inserting effects on the input side, or
dropping in an EQ during recording, which thus processes the recording signal
in real time. These channels are therefore located before the hard disk recording
medium. In the input channels, aux-sends are intentionally left out because they
would make no sense for effects or headphone sound. For example, if reverb were
placed on the vocals being recorded via aux-send here, then the singer would
indeed hear the reverb while recording, but could not hear it when the recorded
material is played back for a punch-in take. Or if the headphone mix were routed
limited in number, there is no longer the necessity of switching between recording mode and mixing mode, as would be the case using a hardware console with
a limited number of channels. Audio channels allow us to create control room
and headphone mixes.
Headphone signal
path in pre-fader
To create a headphone mix, choose the headphone output in the appropriate
channels in the send dialog box, then switch the send path on, and switch to
pre-fader. On the left side of the mixing console, the channel view switch lets you
show the sends of all channels, making it possible to quickly and easily create an
independent headphone mix.
The audio channel has a particular function that refers to a tape machine function; monitor mode. Here you can choose whether to hear the input signal or the
already recorded signal. When recording, it is very important to understand how
this switch works.
On the left side of the console is the
channel view switch. In the above
send-view of the console you can
see the selected headphone signal
paths, which are switched to „Prefader.” The input assignments can
be seen at the top of the channels.
out at this point, then the previously recorded take would not be audible, since
here we only have the input signal.
So that the signals coming through recording channels can be heard in the control room, the necessary number of audio channels must be created and assigned.
In the upper part of the audio channels is a field where a pull-down menu can be
opened with the mouse. Here you can choose the input channels that you have
already configured. To make another parallel to split consoles, the audio channels
are full-fledged tape return channels. Since these channels are equipped and un-
In the Cubase /Nuendo Preferences dialog (File Menu) you will find several settings for Auto Monitoring in the VST submenu. I recommend using the “Tape
Machine Style.” On tape machines, this mode is also called “Auto Input,” which
better describes its function. Since your DAW is simultaneously serving as a mixer and a virtual tape recorder, this function has been introduced into the mixer
GUI. If Tape Machine Mode is selected, then you don’t really have to worry about
the monitor switch any more. It automatically puts a record-ready track into input mode. While the “tape” is in record-ready mode – the track is “hot” – you
hear the singer’s voice as the input signal so that you can talk about the next take.
As soon as the “tape” is running and you are not recording, you (and the singer in
the headphone mix) hear the previously recorded material. As soon as you switch
the tape machine that is already rolling (in other words, your sequencer is running) to record, then the channel switches the singer’s headphone mix to “input”
so that she can hear the recording at the input. The Auto Input or Tape Machine
Mode function is therefore a prerequisite to smooth transitions between monitoring routing when punching in and out of a previously recorded take. The loop
function automatically switches between the two monitor modes, so that you can
concentrate on the performance.
use the VU meter as an input monitor display for the recording level. In the Pre
position, the unaltered level is displayed.
At the right you can see the signal send points for PFL (Pre-Fader Listening) and
AFL (After Fader Listening) switch. This function is only available in Nuendo and
is for uninterrupted monitoring of tracks over a special “listen bus.” This makes it
possible to check a particular signal, while the master bus reproduces the entire
mix (useful for television broadcasts, for example). Differentiating between PFL
and AFL allows you to monitor signals and levels before and after insert effects
and channel level changes.
The middle of the diagram schematically shows the typical order of the channel
functions and various send points. After the phase switch – which changes the
phase of the input signal by 180° (phase reversal) if necessary – comes input trim.
Then come the first six inserts: EQs, compressors, and other insert effects. These
are followed by the channel EQ, which is good for rudimentary sound processing
needs such as a lowcut filter, then the aux-send points (when in Pre-fader mode).
Since the aux signal comes before the channel fader, the level is independent of
the channel fader position and therefore is a good send point for headphones.
The signal is even present when the channel is muted. This way, you can switch
off a track in the control room mix without affecting the headphone mix (this
function can be changed globally in the Cubase/Nuendo Preferences). Inserts 7
and 8 come after level and mute and – because of their position in the signal path
– are good for limiting and dithering. After the pan pot, the signal is sent to the
selected group or output channel.
ReWire Channels are full-featured audio channels whose signal comes from
ReWire applications such as Reason.
Here is a typical structure of an audio channel. The Tape Machine
Mode automatically switches back and forth between the channel input and HD drive during a recording.
Description of the Audio Channel Diagram
The level display on the VU meter can be taken from three different points in the
signal path. At the input, before the channel trim pot (Pre), before the pan (Post
1), and after the pan (Post 2). I recommend Post 2 as the basic setting because you
see what is actually in the mix. The Pre position makes sense when you want to
MIDI Channels do not have an equivalent in the hardware world. Since all of the
MIDI data is already converted to audio in the mix, it makes sense to hide these
VST Instrument Channels also behave like full-featured audio channels. They
receive signals from VST instruments. The VST channels do not need a monitor
Effect channels are full-featured audio channels that serve as effect returns for
effect signals. Up to eight effects can be cascaded together. The send path of the
effect channels are only available to be sent to output busses for adding effects
into headphone mixes.
Group channels are like funnels in which a number of audio, ReWire, and VST
Instrument channels can be brought together. Effect channels cannot be routed
into group channels because of latency issues, but group channels can be routed
into other group channels, if necessary. To set up routing, use the routing window, which can be found in the second pull-down menu on the upper edge of
the mixer window.
The Output or Master bus is the final “funnel” in which all signals come together
to form the end result; the stereo mixdown.
The Audition Bus provides the monitoring possibility of solo-listening to a track
either PFL or AFL, without interrupting the signal path to the output bus. This
function is only important for live recording. Pan position is taken into account
with AFL.
In Nuendo 4, there are more routing possibilities so that group and effect return
tracks can be routed – for example – back to audio channels for re-recording.
This makes sense in film postproduction, where so-called stems (group channels) are brought together in the final mix to keep up to 500 different channels
NDEPENDENT OF GENRE, sound aesthetics and design structure, each
mixing process follows the same set rules. With this toolbox in your luggage, you will reach your target safely.
Because the boundaries of the mixing process between recording, mixing and
mastering have become ever hazier as a result of digital technology and the use
of audio-sequencers, I have a clear definition of the starting point for our journey
through the mixing process.
Similar to the track on a tape machine, all the audio signals are placed sequentially in time order. No effects have been inserted or actuated via a send path, and
all panorama controls are set to center position. In addition, the mixing console
automation is not activated.
Prior to mixing, all MIDI files should be converted into 32-bit audio files. MIDI
file conversion is an important step from the recording to the mixing stage, and it
also generally saves CPU resources.
You also create an archive record of the status quo; you will still be able to use
audio files when the corresponding virtual instruments or external synthesizer
have long ceased to exist or are no longer compatible with your current operating system. To remove some of the finality from this step, you have the option of
opening a folder track that you name “backup” and where you store all MIDI files
Note for Analog Mixers
ost of the time, even when doing a
mixdown on an analog mixing desk, a
DAW or sequencer will be used as a tape machine.
Here you must clearly separate the work processes
done on the computer and the mixing desk. This is
a prerequisite for working efficiently. Since nearly
all functions are available on both the computer
and the mixer, decisions must be made such as
“panning and level automation will be done on the
computer.” This way, you won’t waste time with
contradictory work processes.
In this case, I will assume that we are dealing with
computer-based mixing, but the same rules can be
applied to working with analog mixing desks
that have already been rendered. I like to place this
folder track at the bottom of the project. If necessary, you can access these MIDI files at a later date.
This step is similar to a mixing project where you
receive the material as OMF-Files (Open Media
Framework) or as continuous audio files.
If you want to mix your own production that is
already halfway to a finished mix, you can copy
your project file and remove or set to zero all effects that are of no artistic design value, together
with all pan and level settings.
This makes particular sense if you have lost track
of things in a mix that has evolved over time. At
this stage, you sometimes find reverb devices in
use that are no longer associated with a clearly defined task and whose undefined use can result in
undefined layering. The same applies to other mixing settings, such as panorama
or EQ. If these settings have no context within the strategic plan for the mixing
process as a whole, they are often pointless. The fluid progression from recording
to mixing can thus result in a kind of blindness or rather deafness to the entire
process; the only way to break from that is to make a clean fresh start.
If you have recorded or received the original files with 16 or 24-bit resolution, I
recommend converting the material into the best possible bit depth before beginning the mix.
After you have converted all audio files into 32-bit floating point format (or into
24-bit integer files if this is not possible), you can then begin work.
One thing has not changed. If you received files for a mixing job in the form of
a collection of continuous audio files, the mixing process begins – as it always
has – with systematic listening to each track in order to make sure there are no
editing mistakes. Also, this is when sound quality and loudness consistency is
checked. In the past, the process of erasing mistakes and unwanted passages in
each track was a source of adrenaline, since the luxury of automation muting did
not exist. Each track would be cleaned up manually with the record button. Even
unwanted breathers on the irreplaceable lead vocal track would be erased using
only the ears and instinct.
Fortunately, cleaning up tracks today is much less dangerous, thanks to being
able to work non-destructively. Despite the today‘s possibilities, it is shocking
how many editing mistakes occur, such as badly-cut breathers, rendered clicks
through unclean cuts, and many other surprises. In nearly every case, it is worthwhile to listen to every single track from A to Z. Sometimes I take the opportunity to separate the “s” sounds and reduce their levels manually. And when you
do not know the material, systematically listening through all the tracks in this
way lets you plunge deeply into the song material and get familiar with it. Even in
your own production, editing mistakes resulting in clicks may have escaped your
attention. Highly edited tracks require careful inspection and prudent use of the
auto-crossfade function, found in most DAWs.
While listening to the tracks, you create clarity and order on the screen by creating folder tracks and assigning colors.
Use your own personal preferences, such as drums = off-white, bass = blue,
acoustic guitar = light yellow, and so on. In most programs, you can save this in
the preferences. Generally, I start with drums, then bass, then harmony instruments and then put the vocals down at the lower end of the arrangement window.
Obviously the order is both a question of what instruments are employed and
your own personal preferences.
As soon as the actual mixing process starts, I place the main harmony instruments after the drums and bass, and finally give the lead vocals their spot, so that
I can put everything else (backing vocals, percussion, strings, and whatever else)
in the gaps.
Basically, I create a bed in which to place the vocals. After successfully putting the
vocals in, I deal with the “decoration.”
A very important step is the creation of group tracks (traditionally called “bus”).
Group tracks are a kind of audio funnel, in which individual audio tracks – or
even other groups, if necessary – are combined for being able to control them
To do this, you have to assign a group to each folder (see screenshot) and then
route the audio tracks to the appropriate group. Naturally this only makes sense
if one or more tracks are assigned to each folder. For the sake of clarity, place the
group track underneath its associated audio tracks within the folder track.
Here is an example of track/group organization in Nuendo/Cubase. The number of groups was previously determined. On the upper edge of the mixer window, you can see the pop-up window that
is used to choose the input and output routing. In the illustration, output BUS 1 plus 6 different
groups can be selected.
The track structure is clear in the Nuendo arrange window: In the folder track “Percussion,” the audio tracks are routed in a group track (14). Folders that are not being
worked on are closed. In the percussion group track (14), two marker tracks are visible.
The upper track is typical in Nuendo/Cubase, the lower track consists of empty MIDI
regions. The regions are useful for quickly selecting parts of the song for looping, while
the markers have count-in bar and can be controlled directly through the computer
keyboard or remote. You can select the MIDI regions and loop them with the keyboard
command “P”. Shift-G activates the commands “loop, cycle, play.”
Using this group constellation within the folder track has a number of advantages
during the practical mixing process:
u You can just open the folder track or tracks that you are currently working on
in order to keep the screen layout as clear as possible, particularly in larger
u When dealing with mute and solo switches, you need to be efficient. Does this
sound familiar: you start off with one solo track, you work on and soon have
10 tracks in solo mode, and then you want to listen to two of these 10 tracks in
solo mode? You can make life a lot simpler for yourself if you consistently use
the solo switch only within a particular group and make sure that you’ve previously muted those folders and channels that you do not want to listen to. For
example, if the drum group is the first one I work on, all other folders and all
the tracks within the drum group apart from the first track that I am working
on are switched to mute. Then I work my way through track by track, by demuting the individual channels; with this process I can use the solo function
for what it is designed for; a solo function. Using the solo switch in this way allows you to easily do A/B comparisons between different audio tracks: Track A
is un-muted and track B is selected, but muted. Now you can use the keyboard
shortcut “S” to switch to track B, because selected track B will be switched to
solo, while at the same time track A is muted.
u If you consistently use the audio channel send paths rather than the group
send paths to control send effects such as reverb and delay, you can use the solo
function of the folder track to listen to “wet” solos and the solo function of the
group track to listen to “dry” solos. This makes it very quick and easy to compare the wet and the dry sound of a group. The logic behind this is as follows;
when you switch to the group solo function, all associated audio channels are
also switched to solo mode, but only the send-return paths of the send paths
actually used within the group. Because the send paths of the group track are
not in use, you can now listen to the dry sound. However, if you switch the
folder track to solo, all tracks within that folder (audio tracks and their associated groups) together with all the used send-effects are switched to solo – as
a result, you hear the group with added effects-returns. This function is very
helpful in assessing the use of reverb creating stage depth. By contrast, adding
the effect amount by switching the FX return channels to solo only works satisfactorily as long as there are no other audio tracks to control that particular
effect. Otherwise, you may well hear the drums with reverb, but also the reverb
part of the chorus and of other tracks using this send effect.
Working with groups has many other advantages. As in a jigsaw puzzle, the individual pieces fit together. First, the drums have a nice fat and spatial sound,
then the puzzle piece with the harmonic instruments is nicely “rounded” off by
distributing the instruments appropriately within the panorama and frequency
bands. Then you use the group fader to rejoin the pieces. The subsequent level
control is much easier because of the coherence in the overall setting of the group
blocks. Defining a good balance in loudness, panorama, and sound – for example
20 individual string tracks – would be quite impossible without using sub-mixes
in groups.
The loudness control of the group tracks is also an advantage in controlling of
the master bus level: a simple combination of the group fader allows you a quick
correction of the master bus level in case of overdrive or underdrive. The level of
the master bus fader should remain at 0 dB; the level is controlled with the limiter
only. To do this, use the level and gain reduction display of a good limiter that
only displays a slight activity of the limiter at significant peaks with a working
range of 1 to 2 dB, counting down from 0 dB full scale. This ensures that only
the peaks of the loudest passages are slightly limited, thus protecting the output
from overdrive, while also making sure at the same time that the gentle use of the
limiter does not cause any damage to the transients.
It also allows you to reduce all group levels evenly in order to create headroom
for a group requiring an even greater level within the overall mix despite having
reached the upper stop of its fader setting.
In order to save valuable CPU or DSP resources in view of the limited availability
of good outboard equipment, group tracks are popularly used to insert EQs and
compressors. If 40 chorus tracks with a good EQ or compressor are inserted into
each, they are likely to be too much for some computers. As a result, some of
the best little treasures are found in the inserts to important tracks such as lead
vocals, bass drums, snares, and to the group channels. The use of latent dynamic
The UAD-1 Compressors 1176 LN
and LA-2A are particularly appreciated for group tracks and important
individual tracks.
processing welds the individual events into a unified sound. Particularly popular
are plug-ins that are full of character such the UAD-1176 compressor or the LA2A.
Specific settings for dynamic compressors will be discussed in the chapter about
“dimension 2.” (See page 130.)
It is better to compress frequently and gently rather than rarely and hard. Working
with compressors in groups has a lot of similarities to working with large analog
consoles. Even in the best studios, the number of good compressors is limited.
If there are a lot of individual drum tracks, a good compressor is applied to the
most important tracks - snare and bass drum - and a group compression is then
used to weld the other individual tracks together with the snare and bass drum to
a single entity. The same applies to lead vocals, doubled tracks and adlibs.
It is much easier to create space by using EQs on the spectral level – i.e. the vertical dimension or Dimension 2 – rather than EQs on lots of individual channels.
For example, if the lead vocals require more space in the frequency range for
warmth – i.e. the lower mid-range between 350 and 400 Hz, this can be easily and
effectively accomplished by using group EQs.
A good graphic analyzer in group-solo mode allows you to quickly unmask the
keyboard pads, which create a fuzzy sound in the vocal range through overlaps,
and counteract this with EQ. A graphic analyzer is a valuable tool to supplement
your ears and to “educate” our most precious sensory organ. Correction occurs
particularly frequently in the lower “mid-range misery” range between 120 and
350 Hz. This range is predestined for overlaps and a fuzzy sound, as almost all
instruments are represented in this area. It is also possible to use group EQs to
increase the trebles of a vocal group in order to increase speech recognition and
to convey the impression that the group in question occupies more of a front
position within the mix.
To avoid losing track of things with panning, you need to make clear strategic
decisions. My decision is; I never control panning via groups, but only through
individual channels. Otherwise you run the risk of getting “into a tangle.” The
same applies to automation. Just imagine the following: the individual channel
and panorama settings are all right, and you decide in the final stage of the mix to
automate the group panning for 4 beats.
The next day you notice that a particular event in this range is placed too far to
the left within the panorama. You counteract this in the audio channel and in that
instant the group automation sets it back to the opposite direction. My recommendation is: “Hands off group panorama”!
It is much better to apply basis widening effects to groups than to the whole mix.
If the number of instruments in a particular frequency range is too great and if it
is difficult to differentiate them clearly, stereo expanWhat is Stereo
sion offers an opportunity to create a new dimension
or at the very least widen the horizontal dimension
(Careful, check for mono compatibility!).
The StereoExpander allows you, for example, to swap
high chorus sections and move them to the extreme
left or right. For guitar walls-of-sound like Rammstein
you can widen the wall by transferring some of the
layered guitars to a separate group in order to insert a
stereo spread.
tereo basis widening uses frequencydependent phase-shifting to create more
wideness on the panorama level than would be
normally possible with speakers placed 60° apart.
Every DAW has plug-ins that does this, usually
called something like “stereo spread” or “stereo
expander.” Overuse of this effect results in phase
cancellation when the mix is switched to mono.
Within the framework of a creative and spontaneous mixing process, the following overview acts as a guideline and reminder, which also takes into account
suggestions made in the later chapters:
u Convert files to 32-bit floating point format whenever necessary and possible.
u Convert to audio the MIDI tracks of external expanders and VST instruments.
Use 32-bit floating point format.
u Check each track individually for breathers, editing mistakes, and clicks; clean
u Label, sort and color-code tracks, assign them to group tracks, folders, and
route them.
u Use the group-solo function to control the routing.
u Define your mixing strategy with a panorama sketch (see chapter on dimension 1 – Panorama).
u Mute all folders and tracks, with the exception of the drum folder and the
individual track you are starting with.
u Start building up the rhythmic backbone, starting with the bass drum, followed by the snare drum, making use of panning, EQing, compression, gates
and reverb, until the folder track “Drums” represents a powerful and rounded
u Balance the group levels of all groups edited so far.
u Distribute decorations and additions in a spectrally sensible manner around
the existing basis.
u If a particular event sounds fuzzy, look for a spot within the three dimensions
where it can be clearly heard. If you cannot find the spot either with a good
panning strategy or with EQing or layering, reconsider the reason for having
this particular event where it is, and possibly mute it or look for a better spot
elsewhere in the song.
u Fine-tuning of volumes at extremely loud or quiet listening levels.
u With a little bit of routine and experience, after 3 to 4 hours the mix is 90 percent ready. The static mix provides the starting point for the hard part of the
work that follows.
u In order to shape the final 10 percent, you need a healthy level of perfectionism. But don’t do this by yourself, be honest with yourself and team up with a
colleague in order to generate the dynamics required to reach higher targets.
u Volume Automation for introducing new events.
u Volume Automation for song structure dynamics.
u Panorama and stereo expander automation for clearing up the last remaining
fuzzy spots.
u Carry out further automation.
u Creative fine-tuning to refine details.
u If the bass is part of the drum group, add it to the mix after editing as required.
u Constantly experiment in order to improve individual events that do not yet
sound right.
u Your next step is to build up the instruments that provide harmony and
warmth. Distribute the elements according to their complimentary spectral
properties to the left and to the right in the panorama.
u Set the brickwall limiter in the master section to - 0.3 dB in order to ensure
slight and consistent peak limiting.
u You can perform the final step either in group solo mode or in conjunction
with the “open” drum group.
u Export the master with a bit of lead-in and lead-out and without fade-out in
32-bit floating point resolution.
u Create a good lead vocal sound and add it to the center.
“That’s easy, it’s just a matter of left-right distribution.” How wrong can you
be! If you do not do your homework in this area, you will soon encounter
problems on the way to a tidy mix, because a good plan for the panorama is the
basis of a good mix.
Start the mix by drafting a mixing strategy in which you set out the placement of
Create a mix strategy sketch at the beginning of the mixing process.
individual instruments, almost like placing them on a stage. For the sake of simplicity, we will assume that the instrumentation is consistent and continuous, thus
allowing us a static panorama distribution. Of course, in the distribution, you do
not simply choose randomly, but apply certain rules. The most important aspect
in planning your panorama outline is three-dimensional thinking! Dimension 1
- the panorama distribution - is the horizontal level on your stage plan, you have
to imagine dimension 2 – the frequency distribution - and dimension 3 – depth
- appears as the vertical axis in this two-dimensional representation.
Start with those aspects that have been tried and tested a million times: bass
drums and snares are placed at the center. This is essential for the simple reason that the surfaces of two loudspeaker membranes are better at simultaneously
transmitting a deep sound transfer than a single loudspeaker with half the surface area. That would be the case, for example, if you placed the bass drum to the
left. The same applies to all deep or ultra-deep sound events. Deep sound parts
that are placed at the outer extremes of the sound panorama can easily lead to a
lack of mono-compatibility and a bad energy distribution within the overall appearance of the panorama. The goniometer or a good peak meter will alert you
to an uneven energy distribution within the panorama. Accordingly, the bass is
also placed at the center. The remaining drums are now placed more or less naturally across the panorama, according to the style you wish to achieve. For natural
drums, I always pan from the position of the listener. I visualize the drums in
front of me like a live mixer would see them on stage, and I place the hi-hats a bit
to the right, the high tom a quarter to the right, the mid tom at the center, and the
floor tom further to the left. It is essential during panning to check the overheads
to determine whether the right-hand side is also the hi-hat side. At this stage, always monitor your correlation meter to check that the entire range does not slip
into the red sector as a result of the total left-right spread of the stereo overhead
track.. This should apply to most procedures related to setting up the microphone
and is simply remedied by gently limiting the panorama to a 9 o’clock to 3 o’clock
setting, if you visualize the pan knobs. Now you correlate - and the correlation
meter indicates with a cheerful green light that the tracks harmonize better.
The lead vocals naturally take center stage at the front; now we look at the second
dimension in our mental visualization, because the center is already crowded
with so many events; the bass exclusively occupies the lowest one-and-a-half octaves in the center. This is not a place for any other instrument! Above that is the
bottom sector of the bass drum at about 80 to 100 Hz. The heads create localiza-
tion with the clapping sound; depending on the bass drum sound, it has quite a
broadband presence in the upper mid-range to 6 kHz. Because this section of the
bass drum sound is characterized by transients and occurs in obstinate rhythmic
sequences, this range does not generally interfere with the upper mid-range of
the lead vocals. Above the bass drums are the toms, which we will ignore here
because of the panorama distribution and their sporadic occurrence. The next instrument in the center is the bottom sector of the snares at 120 to 260 Hz. Because
of their rhythmic transient character, the snares coexist peacefully with the lead
vocals without any spectral competition. The main body of the lead vocals sets in
the lower mid or “misery” range, responsible for the warmth of the sound, and
from there develops upwards into the higher regions. The range around 1 kHz is
responsible for the nasal sound, whereas speech comprehensibility for the male
voice begins at about 2.5 kHz, and at slightly above 3 kHz for female voices. Ssounds and similar consonants are located between 6 and 8 kHz, above that the
“shiny” upper notes. The snares broadly occupy the area from the upper midrange to the treble areas. The only range so far with a potential cause for concern
is the lower mid-range, because it is also occupied by the bass drum and the bass.
However, because this range is less important to bass drum and bass, it is possible
to counteract this slightly with parametric filters, resulting in a slight thinning of
the range between 180 and 250 Hz.
Keeping these thoughts in mind, you can now distribute the harmonic instruments across the panorama, which provide warmth and the harmonic basis.
Here, it is always my aim to maintain a complimentary spectral left-right balance.
For example, if there are two harmonic instruments, like a Rhodes and a rhythm
guitar - both played in the lower ranges - in addition to two higher events such
as strings and a synthesizer phrase - you have to distribute the four events in the
panorama in such a way that they do not overlap spectrally – in other words, in
the second dimension. That’s why we place the Rhodes on the left and the guitar
on the right. We also distribute the other two events to the left and to the right
in the panorama in order to achieve a complimentary frequency distribution.
You have to take a fairly radical approach to this in order to ensure that the lead
vocals remain unchallenged in their center position. It is a very easy beginner’s
mistake at this stage to take fright at the existence of so many stereo tracks and
to make very little use of panning as a design aid or to use it too hesitantly. The
fact that the Rhodes is a stereo track and that the stereo chorus sounds simply
great is no excuse for placing it in full stereo width right bang at the center! The
left Rhodes track can also be set at the far left and the right track to 10 o’ clock, if
that is more suitable for the overall picture. This way, there is sufficient space on
the right for the guitar and it now no longer overlaps with the frequency domain
of the Rhodes.
Take the same approach to all other similar matters. If the hi-hats are slightly to
the right in the panorama, place the shaker as a counterweight slightly to the left.
Mixes without an overall panorama strategy frequently suffer from an imbalance in the energy between one channel and the other. The distribution strategy
becomes more complicated, the more instruments and events there are to be distributed and the more inconsistent their appearance. In this situation, I recommend separate panorama outlines for the various parts; if necessary, the guitar
switches sides in the subsequent part, if this is what is required by the instrumentation. In complex pop music, there is no detrimental effect associated with
instruments switching sides within the panorama. However, any unnecessary
panorama switches should be avoided in more intimate music with a consistent
To underpin your panorama strategy, follow these rules:
The acoustic rule that deep sounds spread in circular form and can hardly be detected below 100 Hz, whereas high frequencies spread directionally and are easy
to detect gives rise to the following panning rule:
Choruses, orchestrations and general instruments in pop productions are generally placed in such a way that deeper sounding instruments are always placed
more towards the center; the higher the sound event, the further out they are
likely to be placed. Within a chorus group, you would thus place the high voices
further to the sides in the panorama, and deep voices more towards the center.
The same applies to string ensembles, orchestrations and almost all many-voiced
(thick chorus) sound events.
Of course there are exceptions to this rule, depending on your own preferences
and artistic freedom.
The art of a skillful panorama distribution is to pay due attention to the second
dimension (the spectrum), which is not part of the outline - and that at least
three times as much! The spectral distribution across the center is a slightly variable standard, and, as far as left and right are concerned, it makes sense also to
be aware of the spectral distribution. Distribute the events in such a way that
the frequency distribution is also balanced in the outer sectors. If a mix is very
complex, you can approach this spectral distribution in a series of “slices” such as
“outer left - 9:00 to 10:30 – center – 1:30 to 3:00 – outer right” in order to find a
place for each event. In this instance, using a stereo expander can create an extra
lease take a mono bass track and duplicate this at a 1:1 ratio on a different
track without moving in the timeline. The easiest way to do this is by using
the “Duplicate track” function. Now rotate the phase of the copied track by 180
degrees. Now pan one track on the far left and the other on the far right. Listen
to the result while switching the mono switch of your control room matrix or the
control room mixer on and off. Bingo! If your DAW does not have a mono switch in
the master section, you can use a plug-in that has one or you can put both channel
panning knobs in the center position. When the mono switch is not pressed, you
will hear the acoustic phase interference in the room, which is not as strong due
to natural spatial de-correlation. When the mono switch is pressed, you’ll hear the
physical phase interference, when both channels are mixed together.
Repeat this exercise in WaveLab with stereo files and finished mixes by selecting
one channel and reversing the phase with the command Invert phase in the
Process menu. Here, too, the difference between acoustic and physical phase
interference is very obvious, with the latter occurring whenever your stereo mix
is played on a mono device. Once you have become familiar with what phase
displacements sound like, the regular use of the mono switch is a simple, but
effective device to monitor for unintentional phase interferences.
You can visually check the phase with the correlation meter, which very clearly
shows that the signals with the same frequency on the right and left stereo
channels add up in their amplitudes as long as the phase of both signals is within
the 0-to-90 degrees sector – generally shown as green.
Please refer to the section “Correlation Meter.”
u Before mixing, make a sketch of your panning strategy.
u Anything that is not bass, bass drum, snare or lead vocals should not be in the
u Instruments present in the same or overlaying frequency sectors should be
placed at opposite ends - complimenting each other - within the panorama.
u Once you have established the basic static panning, but still find parts within
the song that are unsatisfactory in their transparency, it is always worth an attempt to remedy this by using panning automation.
u Well planned and carefully automated panning often creates greater clarity
in the mix than the use of EQ and is much better than unnecessary EQing.
If an event is drowned in a sound mush, your first step to solve the problem
should involve the panorama knob before resorting to EQ.
u Used in a controlled fashion, widening the sound basis can create extra space
in the horizontal dimension and thus ensure a clearer sound. Check by using
the correlation meter and the mono switch on your monitoring board. Please
refer to the Sound Examples #06 and #07 to illustrate this.
Sound Example
#06 & #07
u Be courageous! Try extreme panorama settings, making the center free for
lead vocals.
The mono-switch in the master section is a way to control mono-compatibility that is much underused in the native world. Even today, there are still many
mono radios and TVs. If your goal is to get airplay for your production or to
publish the mix on vinyl, please activate the mono switch regularly for checking
for compatibility. If the mix collapses as soon as the mono switch is pressed, then
there are probably too many deep sounds distributed across the panorama or
there are simply too many phase interferences.
Phase Interference Mostly Occurs in the
Following Cases:
u Bad stereophonic recordings: when you record a signal using two microphones, you have to position them in such a way that no phase interference
occurs. For that reason, they are positioned either with the capsules at equal
distance and on the same axis to the source of the sound, or at such a distance
that the distance of the second microphone to the source is at least three times
that of the first. Here, too, during the recording you should monitor the phasing with a correlation meter in combination with a mono switch.
u Simultaneous multi-channel recordings with correlation errors (mostly in the
case of drum or live recordings). For example: a snare recorded with a top microphone and a snare strainer microphone generally has phase interference,
if the phase reversal switch on the snare strainer microphone channel is not
activated. Phase problems between the snare top microphone and overhead
microphones can also lead to interferences. Controlled repositioning of the
microphones and/or the use of the phase reversal switch on the console or
the microphone preamplifier can remedy this. If the recording along with its
errors is already “on tape,” using the phase-shift switch in the individual channels can remedy the situation at the mixing stage. Thanks to HD systems with
sample-exact zoom functions you can also use extreme enlargement of an audio file and move it by just a few samples – and maybe solve the problem that
way. This simulates the change in the distance between microphone and sound
u Keyboard pads and layered sounds often contain phase effects that are not
mono-compatible. Thanks to the three different panning functions available
in Nuendo & Cubase (Stereo Dual Panner, Stereo Combined Panner, Stereo
Balance Panner), you can replace the old panning slider in a stereo track with
separate panners and, use them to limit the stereo width from 9:00 to 3:00 - in
the same way as with classic knobs. The resulting signal is generally still broad
enough, but mono-compatible. The same trick can be used for problematic
overhead tracks recorded with microphones not positioned on an axis. If your
DAW only has a panning slider in the stereo channels, I suggest using small
plug-ins. For example, the PSP Stereo Controller from the PSP Stereo Pack
lets you easily adjust stereo width. Logic has separate pan controls on each
individual L-R channel.
u Reverb returns; the same applies. The 9:00/3:00 setting - or the 80 left/80 right
setting, to talk in Nuendo readings - generally remedies the situation. But be
careful: limiting the width is not always necessary and can create undesirable
masking. (The masking effect will be explained later.)
to create more space at certain points, insert a stereo basis widening effect into
the group in question, which remains in a neutral position throughout the song
and, using automation, can be targeted at these specific fuzzy locations to create
more space.
u Excessive use of stereo spread; use a lower setting.
The Masking Effect
u Chorus and other phase effects; if you really want to use a chorus on a bass,
select one which only allows you to process frequencies above 250 Hz. If that is
not available, copy the track, apply a steep lowcut filter at 250 Hz to the copied
track and only use a chorus on this copy. Another possibility is TC’s FX matrix
called “FX Machine,” which has the option of splitting a single audio source
with a crossover, so that the higher frequency range is processed separately.
The FX Machine is discontinued.
The masking effect describes the phenomenon where a high-amplitude frequency
masks a neighboring frequency of lower amplitude. This principle is also familiar
from audio data reduction, like the MP3 compression format. Being aware of this
effect helps with the mixing process. But first, we should describe the masking
effect more precisely.
u Unintentional shifting of tracks during editing. If phase interferences occur
consistently, it is worthwhile to check the offset between the two channels.
Stereo width and mono-compatibility are contradictory.
The actual stereo width - depending on the target format (for example, CD, TV,
or radio) – is a matter for the engineer to consider. The correlation meter helps
in the decision making process, and the mono switch provides a way of checking to support the decision. Not every flicker into the red area of the correlation
meter necessarily requires corrective action. Such action should only be taken if
important information is lost as a result of switching to mono. The width setting
of the stereo expander should only be adjusted while using the mono switch for
constant monitoring.
To conclude this topic of panorama control, I will describe a situation during a
mix; due to legal reasons, I cannot include this on the accompanying DVD:
The lead vocals were embedded in a chorus in the refrain. The chorus had a leading function for the hook. For one bar, the lead vocals transgressed from the hook
line, causing an unpleasant interference. For this one bar, the mix sounded mushy
and fuzzy. Neither the chorus nor the lead vocal track was clear and easy to isolate.
With the aid of a simple panning automation, which I used to place the chorus
into the entire left-right width for this one bar, I could solve the entire problem.
If you find in such a situation that, the normal stereo width is insufficient in order
The more congruence there is between the direction from where two different signals of similar frequency originate, the stronger the masking effect on the quieter
signal, and vice versa. In nature, there would be no masking whatsoever, as long
as these two signals of similar frequency originated from completely different directions. This lets us conclude that the masking effect is much more pronounced
when we have fewer possibilities to separate frequencies in the form of discrete
audio channels. To put it simply, in a mono reproduction, all sound originates
from just one direction and as a result, the mix suffers from the masking effect,
whereas the masking effect has little impact in a surround sound reproduction, as
long as the mixing engineer has taken it into account in the general panorama distribution. Accordingly, you can deduce the rule that mixes with a lower number
of reproduction channels contain less spatial information, whereas a surround
sound mix can handle a lot of spatial information without becoming fuzzy.
The masking problem is already taken into consideration, assuming that you
have a clear panning plan which involves left-center-right frequency distribution,
as suggested in the earlier section on panorama strategies. Anyone professionally
involved with sound knows the effect in conjunction with the application of reverb. If you apply reverb to a lead vocal during a long-held note, the reverb, which
consists of a very similar frequency structure, only becomes audible once this
note has become much quieter or has finished altogether. We can make direct use
of this insight for the application of reverb to stereo and surround sound mixes.
If the reverb response of a center signal tends to be masked in the case of a stereo mix, try a spatial separation of the reverb section from the center by using a
stereo expander. Use the mono switch to check phasing and mono-compatibility,
in order to achieve a happy medium between transparency and mono-compatibility.
Signals that are placed on the left hand side within the static panorama can be
used with reverb that is weighted to the right, and vice versa. In that way, the listener is provided with the spatial information without watering down the mix.
In surround sound mixes, you pan the stereo reverb return for signals that are
oriented towards the front to surround left and surround right, and the reverb
return for signals placed at SL and SR to left and right.
Don’t worry – decoupling direct signals from reverb won’t irritate the listener.
Imagine the following real-life situation; you stand in the front third of a large
church, looking towards the altar, and you are addressed loudly from the altar
from a few meters away. In addition to the direct signal, the full reverb tail spreads
forward from the back of the church to the front. Would that make you think that
you were being addressed from the back?
Let’s summarize the salient points with respect to masking:
u Because of directional masking, mixes that are supposed to be mono-compatible – for example for TV – can handle less reverb than mixes that are predominantly intended for stereo or even surround sound productions. Mono
mixes thus tend to become “mushy” or over saturated much earlier, because
the reverb increasingly dilutes the perception of the direct sound.
u The greater the number of available loudspeakers for the reproduction of a
recording, the greater the amount of reverb that can be used. In order to put
into place a clear spatial structure without over saturating the mix with spatial
information, you should keep the reverb away from the direction that the dry
sound originates from. This avoids masking.
As far as I know, Nuendo, Cubase, and Cakewalk SONAR 6.0 are the only sequencers that have a stereo pan mode with an adjustable panning behavior.
In Sonar, there are six different panorama characteristics available, whereas with
the Steinberg products, there are five modes available in the Project Setup dialog
“Stereo Pan Law” that provide automatic level adjustment for mono tracks. This
is based on the observation that a mono signal which is reproduced simultaneously and at equal volume on the left and on the right, i.e. at the center, sounds
louder than it would if was only reproduced at the far right or at the far left. The
default setting for the automatic level adjuster is minus 3 dB and results in a level
reduction of about 3 dB in the center position. At this setting, a mono signal
always retains the same physical sound level regardless of its position within the
panorama. The settings minus 6 dB and minus 4.5 dB reduce the mono signal at
the center correspondingly.
At the 0 dB setting, there is no level compensation, and right-left channels brought
together in the center are twice as loud as they would be if they were positioned
fully left and right. These different so-called “panning laws” correspond to the
behavior of analog mixing desks.
This function is very useful for eliminating the need for level adjustments after
panning changes. Experiment to find the right settings for your needs. In my
experience, the settings between 3 and 6 dB are particularly useful when you mix
orchestral music for mono-compatible TV productions.
u Masking also applies to different sounds that occupy similar frequency ranges.
That is why a transparent mix requires that instruments and events with overlapping frequencies be placed in complimentary positions across the panorama, according to the panorama strategy made at the beginning of your mixing
HE SECOND DIMENSION concerns skillfully distributing events in the
frequency spectrum. We have already accomplished the very important
step of distributing events in the panorama (as described in the previous
chapter) by taking the frequency characteristics of individual instruments within
the left-center-right range into account. This now raises the question of purism.
Purism is only appropriate when no corners have been cut at the front-end and
the recording engineer or producer knew exactly which sound he or she wanted
to hear. If the recorded audio files are delivered in perfect shape, it suffices to use
EQs sparingly, especially in the lowcut range, in order to remove DC offsets and
low-frequency artifacts such as breathers, pops, and so on. For our purposes, we
will assume a “worst case scenario” to cover all situations.
As I have already mentioned in the section on studio equipment, channel EQs are
useful for rough work and “external” DSP-aided EQs, and particularly good native plug-ins are best for fine tuning. We require devices with a particular warm
and natural sound with very little tendency towards unnatural filter sounds. They
also need to be precisely adjustable.
The differences and limitations of native and DSP-aided EQs are most obvious
when these devices have to cope with significant increases between 14 and 20
kHz. You should take the trouble to compare your EQs with each other and also
possibly with quality external devices. You will be surprised at the differences.
Especially in the higher regions, even a mediocre tube EQ simulation can sound
harsh, digital, and quite simply, awful. A further indication of the quality of an
EQ is its sound behavior at extreme bell filter settings. Comparisons will help you
to achieve better results, because you will then be able to restrict yourself to a few
well-chosen tools.
For precise operations of a tonal nature, I have prepared a frequency chart. It
shows the relationship between equal-tempered notes and their corresponding frequencies. The list is also available in pdf-format in the data section of the
DVD-ROM so that you can print the file out for quick reference.
Frequency Chart
The low B at 30.94 Hz equates to the open B string of a 5-string bass.
In order to organize the spectral distribution of our sound events, we first need an
overview of the relevant frequency range from 1 Hz up to about 22 kHz.
Description of the Frequency Bands
u The sub-bass sector – occasionally also referred to as infra-sector – covers the range
from 0 Hz. to approx. 25 Hz. The frequency list shows you that for all practical
purposes there are no musical sound events in this range.
Mixes where lowcut has been inconsistently used to prevent low interference signals
contain artifacts in this range - popping sounds, rumble, and many other interfering
signals. A frequency component between 0 and 1 Hz is called DC offset and also
must be eliminated. Consistently cleaning the sub-bass range is a prerequisite for a
good bass sound. My rule of thumb for lowcuts is: “a good mix requires at least the
same number of lowcut filters as there are tracks.”
You will find more information on this topic in the section “How do I mix a
tight bass range?”
A good mix
requires at
least the same
number of
lowcut filters as
there are tracks.
u The bass sector proper comprises the range from 25 Hz to 120 Hz: the
bass sector is first and foremost the sector for the bass. That might sound
trite, but it is important! The frequency list shows you that the deepest
note B of a 5-string bass has a frequency of 30.94 Hz. If your bass drum
has a characteristic core frequency of 90 Hz, this means that there are
almost 1.5 octaves available for the bass alone. If the bass drum is not
too low and if a lowcut filter is used only sparingly, there is hardly any
interference, and the range provides enough scope to manipulate the
bass independently – even in the finished mix. Sometimes you can hear
bass dead spots. These are individual notes played on imperfectly made
instruments, which sound quieter than other notes because of instrument construction errors. With the aid of the frequency list, you can equalize these dead
spots accurately by accessing them precisely using a parametric EQ with steep
correction in order to make sure that surrounding notes are unaffected.
The bass sector includes the frequency range of the bass drum and ends at
about 120 Hz. Together with the next higher frequency sector, the bass sector
is crucial for the perception of “warmth” in a particular recording.
The bass drum should be somewhere between 75 and 100 Hz. For club music,
you should consider that depth is not the same as pressure. In clubs, pressure develops at around 90 Hz, because the system cannot powerfully transmit
bass drums that are much lower. The lower a bass drum, the more difficult
it is to edit. Bass drums that extend to 60 Hz or deeper have to be carefully
adjusted to the key of the song. Deep bass drums tend to have their own tonality, which runs the risk of interfering with the bass, and causing the sound to
appear washed out. Another rule for dealing with very deep sound events is:
“the deeper the shorter!” Cut deep events short so that they do not take up too
much space within the mix.
u The lower mid-range “misery” or low mid range sector between 120 and about
350 Hz is the second pillar for the “warmth” in a song, but also a potential
source of unpleasant misery. You should pay particular attention to this frequency band during mixing, because almost all instruments are present here.
Stick consistently to the mixing strategy that you have outlined at the start. The
central part of this section belongs to the lead vocals, and the left and the right
to the harmonic instruments providing the warmth. It is better to thin out all
other events in this area rather than risk overlaps. This is the section where one
finds most of the acoustic recording errors, for example when recording lead
vocals, too little attention is given to the comb filter effects of room acoustics.
Room modes in particular affect this recording range and add up to produce
tedious misery. Should you find this to be the case, use notch or parametric
filters to look for the problem frequencies in the affected tracks, and lower
them steeply. However, you need to ensure a good balance between warmth
and muddiness. If you deliver a mix where these frequencies were imprecisely
and therefore unsuccessfully eliminated – taking away much warmth while
still being muddy – the mastering studio can do little to remedy the situation,
because the mastering engineer cannot improve one feature without worsening another in this case.
u It is very difficult to draw up a set of rules for the nasal mid range of between
350 and 2000 Hz. Any rectifications in this area have to be done very much on
a case-by-case basis. If too much emphasis is placed on this range, the resulting
sound is nasal, woody, and piercing.
u The upper mid range between 2 kHz and 8 kHz is responsible for speech comprehensibility! As a basic principle, please check whether a broadband lift on
the lead vocal track between 2.5 and 4 kHz would make sense in order to aid
comprehensibility. The drum skins, which are important for the rhythmic localization of bass drums, are also present in this range.
u The treble range covers the frequencies from 8 kHz to 12 kHz. This is home for
cymbals, high percussion instruments, S-sounds, chimes, and the high range
of many instruments.
Artifacts &
DC Offset
Bass & Bass Drum
around 1 kHz)
Speech Comprehensibility
(Particularly between 2,5 and 4kHz)
Air Band
Sibilance (8kHz and higher)
Upper Trebles
Sub Bass
> 25 Hz
25 Hz – 120 Hz
Lower Mid-Range
120 Hz – 350 Hz
350 Hz – 2kHz
Upper Mid-Range
2 kHz – 8kHz
The various bands of the frequency spectrum
u The upper trebles between 12 kHz and 22 kHz are often referred to as
air band. A broadband lift in this range can aid the airiness of a recording. However, an overemphasis can result in a digital or harsh perception.
For a natural sound in the upper spectrum, the trebles from about 12 kHz onwards gradually decrease in loudness, called roll-off. The following illustration
shows the roll-off in the high end of the frequency spectrum.
Before we concentrate on the actual use of EQs, we will look more closely at the
various filter types.
The Most Important Filter Types:
u The parametric filter is generally a bell filter with three adjustable parameters;
Gain increase or reduction, characteristic core frequency (F) and quality (Q).
Quality – also referred to as Q-factor- is given as a mathematically calculated
factor; higher values up to 20 denote greater edge steepness, whereas lower
values up to 0.1 a greater filter width. In traditional parametric filters in the analog world, for example, the ones built into Neve consoles, there is interdependence between gain and Q due to the circuitry, with Q decreasing – in other
words getting broader – when gain increases. This feature is often referred to
as “musical,” whereas the decoupling of these parameters is perceived as cool
and clinical. SSL consoles of the 4000 series are typical representatives of this
EQ characteristic. The Sonnox Oxford EQ and the SSL Duende Channel Strip
EQ offer a choice of both filter characteristics. SPL’s PQ-Mastering is the first
and only hardware EQ that can be switched between these two filter types.
u The shelving filter lowers or respectively amplifies the entire frequency band
above or below frequency F. Depending on the type of shelving, deflections
occur in the lower or upper transitional area - also referred to as overshoots,
the characteristic feature of this kind of filter. The UAD-1 Cambridge EQ for
example reproduces three standard shelving filter versions.
Working with EQs
Parametric EQs are the type most commonly used in mixing. My favorites are the
Cambridge and Precision EQ for the UAD-1, the SSL Duende Channel Strip EQ,
and the Sonnox Oxford-EQ. Unfortunately, the Precision EQ only has grid positions for the frequency selection, which can be somewhat imprecise for certain
cases, but it excels in particularly soft high frequencies thanks to fourfold oversampling. The EQ shows how the plug-in’s discrete oversampling – in other words,
multiplication of the sample rate – can produce an outstandingly soft sound in
the upper frequency bands, without the hassle of dealing with high sample rates
throughout your project.
u Conversely, a lowpass filter, which lets frequencies below F pass, equates to a
highcut filter.
I have often observed that a suitable EQ is picked out and inserted while a loop is
playing non-stop. This numbs our hearing, because our ears quickly become accustomed to acoustic defects. This can result in the engineer to becoming accustomed to the imperfections before the right plug-in is found and adjusted properly. My tip for the use of EQs is to keep a mental image of your spontaneous sound
impression: stop the playback, select the plug-in, and make the settings quasi
blindly according to your ideas. Then you can check on the results with “fresh
ears” and make any finer adjustments that may be necessary. For the targeted
removal of annoying frequency ranges, it is quite legitimate to use a parametric
filter with a high Q-factor and strong amplification, and to sweep forwards towards the problem area. That way, you can target problem frequencies exactly
and dampen them accurately.
u The bandpass filter only lets a certain bandwidth of frequencies pass and is
used in graphic EQs, de-noisers, and crossovers. It consists of a series of highpass and lowpass filters.
To achieve the ultimate target of a balanced frequency distribution, equalizers are
used in mixing to equalize and distort individual audio, group, and master bus
tracks. The basic prerequisites for appropriately using EQs in mixing are:
u The Bessel filter is used as a lowcut or a highcut filter; its effect extends far into
the passband. For that reason, it is unsuitable for hard cuts; one example is the
effective removal of sub-bass artifacts.
u A clear and logical panorama mixing concept, taking into account a balanced
frequency distribution across the Left-Center-Right range.
u The notch filter is very steep and is particularly suitable for the radical removal
of interfering frequencies. The notch filter has to be applied with great precision, because it is not adjustable as to the level of the reduction; inaccurate settings can damage the sound due to the high level of reduction. The adjustable
version is a parametric filter with a high Q-factor.
u The highpass filter allows frequencies above the cut-off frequency to pass and
thus equates to a lowcut filter blocking frequencies below F depending on its
u The Butterworth filter is better suited as a lowcut filter because of its characteristic feature of a steep slope at the cutoff frequency.
u A good understanding of the frequency ranges, where each instrument can
fulfill its role. Many instruments are effective in two crucial frequency areas
while other instruments only operate within a single frequency band.
I will go more deeply into this in Chapter 9, “Working with Individual
Keeping these two conditions in mind, use EQs above all to thin out and remove
interferences and artifacts – lowcut for the sub-bass sector, for example – or to
thin out and create space in frequency areas that are of little importance for the
instrument in question and, only as a second step, use them to emphasize frequency areas that have been underemphasized at the recording stage.
First Remove and Then Add!
A Good Mix Starts with the Arrangement!
Depending on the complexity of the mix, it is important to split the frequency
range three-ways into left - center - right to achieve transparency.
Of course, you should only lower those frequencies of an instrument that are of
lesser significance for its sound character and conversely emphasize only those
that are essential.
There are some basic practices in the use of Q-factors; decreases can be done
steeply with a high Q-factor in order to lower problem frequency areas without
affecting neighboring areas more than absolutely necessary, whereas increases
are done rather more broadband, i.e. with a low Q-factor. This is because high
Q-factors have a tendency to produce filter sounds when raised. Such manipulations are easily detected and perceived as artificial. Exceptions to this include
weak-sounding toms, which can be brought to life with a good filter.
It is impossible for each soloed instrument or group to always have an outstanding sound on its own. However, with a skillful mixing concept combined with
tight instrumentation and a good recording, it is definitely possible for each soloswitched instrument and the total mix to sound very good.
This does not necessarily apply to pop productions. An acoustic guitar that beautifully blends into a pop mix often sounds pitifully thin when switched to solo.
The same can be true with chorus groups. In those cases, it is a good idea to
start off by mixing a harmonious overall sound for the group in quesA good tion and then apply EQing to the group within the framework of your
arranger works overall mixing strategy. If there are too many harmonic instruments
competing for space in the lower midrange on the left and the right
with frequency of the panorama, less important instruments can survive significant
distribution reductions in sound and still find an adequate place in the overall mix
without resulting in an overlapping or fuzzy sound. This could apply,
in mind. for example, to an organ, which might lose its lower mid-range, but
places additional emphasis on the upper mid range with its tinny distortions.
Remember, during the recording and production stage, each additional instrument raises the demands made on the mix. A good arranger works with frequency distribution in mind.
Within the panorama dimension, the vertical frequency dimension should be
split into at least three different ranges (L-C-R).
I will illustrate this approach using a typical and quite common example. As already explained in the earlier chapter on panorama, the center section is the home
ground of the bass, the bass drum, the snare, and the lead vocals. Any overlaps
in the lower mid-range in which all events are present could already at this stage
lead to overlapping and fighting for space. With the knowledge that the lower
mid-range is of no importance for the bass or the bass drums, it is very simple
to create space in this area for the lead vocals. The bass gains its warmth from
the lower one and a half to two octaves, and its tonal comprehensibility between
400 and 800 Hz. The range between 800 and 1200 Hz makes the bass sound
more woody and “planky.” A Musicman-Stingray only works to about 800 Hz,
whereas a jazz bass can go as far as 1500 Hz. The clicking sounds of the frets as
well as slapping are of course higher. The bottom of the bass drum is at about 90
Hz and its rhythmic localization in the upper mid-range. Correspondingly, the
lower mid-range is rather unimportant for the sound and can be reduced. This
results in a balanced sound in the distribution of the instruments’ characteristic
core frequencies. Apply the same principle to the frequency distribution in the
left and in the right section of the panorama. For very complex mixes, imagine
the frequency distribution as split into five sections in order to find space for each
Mixing a Tight Bass Range
One of the distinguishing features of a good mix is that it can be listened to at a
much higher volume than a bad mix without getting close to the clipping limit of
your monitors. The bass sector has been cleaned of all unnecessary information
and artifacts, and the energy reserves are therefore available for essential information. For example, if a recording contains a loud interfering sound at 16 Hz,
which we can hardly hear with the monitors, it will still literally gobbles up the
energy of the amplifier and speaker. This results in a spongy signal reproduction
and speakers clipping, even though the actual audible level is relatively low. The
more consistently I clean up and use lowcuts, the better my bass sound becomes
without ever actually making any changes to the bass sound itself. My suggestion
to use at least as many lowcuts as tracks means; for a vocal track, use the lowcut in
the microphone, the lowcut in the microphone preamplifier, the lowcut in the inserted EQ of the lead vocal track, and a lowcut in the corresponding vocal group.
That is four lowcuts for one track. Almost all acoustically recorded sounds can
contain low frequency interferences. In high string samples, even expensive sample libraries can produce extreme artifacts in the sub-bass sector, which are hardly detectable even with large monitors, but still have a noticeable negative
impact on the sound. Analog synthesizers also tend to produce such artifacts. For a punchy bass drum, it is sometimes also helpful to apply a very
steep lowcut at 50 Hz in order to create more space for the bass. Acoustic
guitars whose main purpose is to create a silky strumming sound can by cut with
a soft lowcut filter (for example at 12 dB/ octave) at 250 Hz. I sometimes also filter
choruses this much or even more. The more disciplined you are at keeping the
bass sector free, the easier it will be for you to mix a full bass sound.
The deeper
the shorter!
If you need very deep notes or a very deep bass drum, remember the rule: The
deeper the shorter! In order to keep these energy-intensive sound events from
preventing the speakers from doing their work; it’s a good idea to keep very deep
sound events short. It becomes very difficult to create a clear mix if the bass drum
has long sustain at about 50 Hz. Such bass drums also tend to blur the tonality
of your song. If you really want to use deep and long bass drums, pitch the bass
drum to the respective chord roots of the song. Listen for example to the CD
Futuresex / Lovesounds by Justin Timberlake.
On the DVD there is an exercise called “Editing Loops,” a practical example of dealing with low bass drums.
EQ Basic Rules:
u Use at least a three-level frequency distribution concept (L-C-R).
u Get to know the sweet-spot frequencies of different instruments.
u First lower, then raise.
u Lower rather steeply, raise broadband.
u Use at least as many lowcut filters as your project has tracks.
u Almost any change in one band will affect the sound in the other bands.
The level of individual events is the second aspect having a substantial impact
on frequency distribution. As with a jigsaw puzzle, one piece determines how
another will fit. The better the preparatory work, the easier the fine adjustments
of the levels. Working with groups is very advantageous because then you only
have to balance the group levels. To create an overall balance between channel,
group and master bus levels or output bus, keep the master bus level at 0 dB and
use a brick wall limiter to protect the output from occasional peaks. Suitable tools
for this are the UAD-1 Precision Limiter, the Powercore-MD3-Brickwall-Limiter
or the Sonnox Limiter.
The actual limiting should only kick in from time to time for occasional peaks.
That’s why it requires constant monitoring to make sure that a high input level
(as a result of groups that are too loud) does not make the limiter work too much.
The limiter should be purely technical in nature, protecting the output from overdrive as a result of occasional peaks. Too much limiting could affect the transients
and negatively affect the sound. If this is the case, reduce the group channel level.
All DAWs have a fader-linking feature so that you can adjust all group faders
together, if necessary. Internal overs, in other words between channels, groups
and master bus, are hardly possible, provided you work consistently with 32-bit
resolution. As previously discussed in the section on bit depth, even levels above
0 dB can be processed perfectly well in 32-bit floating point resolution.
The level adjustment itself depends of course on the genre and personal taste.
Should the vocals be in front as in radio-compatible pop productions? Is it necessary to understand every word? Should the final result be a dance floor mix with
greater emphasis on the bass sector? Different chefs use different recipes, but all
with good results. For this reason, my personal approach is just a suggestion; I
start with the rhythmic backbone (drums and bass), than adding the harmonic
instruments, which provide the foundation to the rhythm group. Then I set the
level for the lead vocal group. After that I add the accessories (percussion, effects,
fillers, and less important instruments). A good trick is to listen to the mix at an
extremely low volume. If you can still recognize the melody, bass drum, snare
and bass, you have achieved initial cohesion.
To determine the relation between snares and bass drums, an RTW peak meter
or its software equivalent is a useful tool. Depending on the style, kick and snare
should usually have the same level of energy. The graphic analyzer can also be
useful in setting the levels. At the very least, it can save you from
with very
serious misjudgments.
Don´t start
before getting
ready with a
static mix!
If there are uncertainties regarding the levels for certain tracks, it helps to
pull the faders in question down and then to bring them slowly up again.
The volume level of the tracks should be so consistent that it is possible
to achieve a good transparent result without the use of fader automation.
The static mix is the starting point for fine tuning with fader automation.
If you start with automation, before the static mix is coherent, you risk being inefficient and having to rework the automation repeatedly, leading to
confusion (see section on Automation). Inconsistencies are better fixed directly on
the audio takes, events, or clips (or however they are called in your DAW). Many
programs have keyboard shortcuts which allow you to process level changes in
individual files with off-line processing.
time line – release – is a further reason to assign compression to the second dimension. Through the attack parameter, the editing of transients with compressors affects the perception and therefore the level. These are the most important
sound editing features of compressors.
Since I have repeatedly seen that even audio professionals with a great deal of
experience in working with digital or analog signals have problems with the
proper application and operation of compressors, I will discuss compression in
particular detail. There are a number of approaches; in my workshops, I have
found that carefully listening to various ways of using compressors is absolutely
crucial for understanding them. This is why I suggest you load and work through
the sound examples in the Compression Exercises on the DVD into your DAW.
The examples are also provided with compression already added, so that you can
compare your results with mine. Of course, your results will depend quite a bit on
what plug-in you are using. You might take the opportunity to reorganize your
plug-in folder. Instructions for the exercises can be found in the “Compression
Exercises” section below.
At the heart of each compressor is the VCA – the voltage controlled amplifier.
The VCA is controlled with a number of different parameters and readjusts the
amplitude of the input signal – in other words, its volume. In a simplified view,
imagine a little person who controls the volume setting and makes quick up and
down adjustments in correspondence with the input signal in order to shape the
amplitude of the signal.
Common Basic Parameters of Compressors and
Their Use
A compressor is a “thickener” which reduces the dynamics of a sound event and
therefore initially also its volume. The actual gain in volume is achieved by returning the level of the compressed quieter signal after compression back to the
initial level or even increasing it. This parameter for automatic level adjustment is
generally called “Auto-Makeup-Gain.” If that is not available, there is an output
level control “Volume,” “Output” or “Makeup-Gain,” allowing you to increase
the level manually.
I associate compression with the vertical dimension, that of frequency distribution. This is because compression has an effect on the density – and therefore on
the perceived loudness of a sound event. The modulation of basic material on the
The simplest graphic illustration showing how a compressor works is an X/Y
graph, where the vertical axis represents the output level and the horizontal one
the input level. If the output level is the same as the input level – in this case the
compressor is inactive – the resulting graph is a diagonal.
Threshold: The threshold parameter represents the outset or the lower limit
where the compressor begins working. This value is set in decibels; starting from
that value, the compressor begins to process the input signal. Since all level measurements in the digital domain are negative (0 dB = full scale), the threshold
value is given as a negative, in other words minus X dB. The higher this negative
value or respectively the lower the level, the sooner the compressor will start to
work. Lower levels or respectively greater negative values represent greater compression. In the illustrations at left, the threshold levels are set at minus 6 and
minus 9 dB.
If there is no threshold parameter – for example as is the case with the Urei-1176
– use the input level to calibrate the threshold. The higher the input level, the
greater the compression.
The threshold value corresponds to the value of the input signal level. If you want
to compress a snare whose highest peaks extend to 0 dB and select a threshold
value of - 6 dB, you significantly affect the dynamics. At the same setting, and
with an input signal that is 6 dB quieter, the compressor range does not even
Level without compression: output level = input level
Compressor with a threshold of -6 dB
Compressor with a threshold of -9 dB
With a 0 dB peak on the snare and a threshold of -6 dB, the compressor affects the dynamics starting
from the level indicated by the red line.
come close to the range that you actually work in, because even the loudest parts
of the signal might only just reach the threshold value of -6 dB, but not exceed it.
When I explain all the other parameters, you will learn how to interpret the “Gain
reduction display” for a quick adjustment of the threshold value.
The second most important parameter is Ratio. Ratio defines how much the output level is to be reduced once the threshold level has been exceeded. The higher
the value, the greater the compression. The setting “Infinity: 1” equates to limiting and is represented by a characteristic horizontal line once the threshold has
been exceeded.
familiar with this parameter, I recommend choosing some
practice samples similar to the ones used in the following
Compression Exercises, and to play around with the attack
time. The higher the value, the greater the number of artificial transients that are added. The attack time refers to
the time between the crossing of the threshold value and
the beginning of the level reduction. This time offset allows
peaks to get through the compressor that are then, once the
attack period is over, processed according to the ratio settings. The waveform representation lets you actually see the
emphasized transients from the examples.
There are two ways of using the attack time creatively: the
purpose is either to find short unobtrusive settings that
make compression discrete and subtle, or long attack times
to emphasize clearly audible artificial transients.
Compressor with a ratio of 2:1 and a threshold of
-6 dB
The input signal level is changed by a ratio of 2:1
compared to the output signal. If the input signal
(horizontal axis) increases by 6 dB, the output signal (vertical axis) will increase by only 3 dB.
Attack time
in ms
Compressor with
long attack time
and strong artificial
Compressor with a ratio of 20:1 and a threshold of
-6 dB = near-limited at -6 dB
Uncompressed bass
In mastering, settings between 2:1 and a maximum of 4:1 are commonly used.
Very rarely are higher ratios such as 12:1 used for individual instruments, which
is quite extreme. Until you have developed a sure instinct in dealing with compressors, you should stick to settings no higher than 8:1. The higher the ratio
setting, the more limiting character you receive and thus, the more noticeable the
dynamic intervention. Low values result in subtle compression.
The Attack parameter allows you to emphasize the transients and therefore to
increase the “snappiness” – or percussiveness – of the recording. In order to get
With a long attack time, the transients are easy to see.
Long attack times give snares, toms and all other percussive sound events greater
assertiveness, bite, and rhythmic definition. This treatment is also well suited for
funky rhythmic Stratocaster style guitars in order to create their typical snappiness. The Urei-1176 is particularly used for editing vocals, because with a long
attack time it adds an unmistakable bite to the vocals; this is particularly evident
in percussive sounds and hard consonants. Many legendary lead vocal sounds
owe their success to the use of the 1176. Caution, the attack time works the wrong
way round; knob to the left equals long attack time, and knob to the right equals
short attack time!
In some situations, a compressor will have only a subtle effect on the sound, but
within the context of the whole mix, it might give
the individual events more punch and greater loTransients
calizability. Long attack times are always welcome
ransients play a huge role in defining natural
when events sound tired and powerless, or when
sounds for our ear, for example with a
an instrument – such as bass – is difficult to rhythbowed string. With layered sounds that have
mically localize in the mix.
no attack, there are no transients. The more
percussive a sound is, the greater the importance
of the transients. Using too much de-clicking
or de-crackling can negatively affect transients,
damaging and deforming the natural sound of
instruments. This is why it is important to pay
attention to percussive elements in the recorded
material when using such tools. Plug-ins such as
the Sony Oxford Transient Modulator specializes
in processing the transient parts in a signal and is
therefore able to increase or reduce the snappiness,
liveliness, and percussive sound of a piece.
Release is the most important parameter in order to keep the pumping sound as low as possible.
Release refers to the time period that passes until
the amplifier level is back at its initial setting. Once
the signal has exceeded the threshold, the level is
reduced in accordance with the ratio setting. Until
the next time that the threshold is exceeded, the
reduction moves back to the initial setting by increasing the level within the release time. The actual shape of this curve – linear, logarithmic, exponential and so on – determines the character of
a compressor. The ability of a compressor to condense loudness by increasing the quiet parts of the signal through the release of
the amplifier is determined by the shape of the release curve.
If you want to achieve subtle results, adjust the release time to the tempo of the
song. To get an idea of what pumping sounds like, take a song and set the compressor to a low threshold, set a ratio of 20:1 and a fast release time. As a result, the
dynamics will rise and fall in rather unnatural waves called “pumping.”
The snare drum example in the Compression Exercises gives you a better idea of
how to use this parameter. The raspy snare sound can’t be heard very well here. A
shorter release time means that the snare “strainer” sounds – in other words, the
quieter parts of the snare sound – are increased in volume, because the amplifier
springs backs quickly to its original value and thus increases the quieter parts of
the material. Conversely, you can use a long release time to “extend” an acoustic
guitar, which fades too quickly due to a lack of sustain. This technique can also
be used to move bass notes to the foreground that otherwise fade too quickly. If
the compressor makes unwanted sounds – such as headphone crosstalk – too
audible, then try to increase the release time.
There are therefore three ways of using compressor release time:
u Discrete compression settings depending on song tempo and program material.
u Increasing the volume of quiet sections that immediately follow transients like
snare drum strainer sounds. This requires short release times.
u Increasing the volume of quiet sections that would otherwise be drowned
in the mix due to a lack of sustain. With a decreasing input signal level, the
VCA (Voltage Controlled Amplifier) jumps back to the starting position, thus
amplifying the weakening signal. In that way, even a mediocre bass with old
strings gets sufficient sustain for a good foundation. This requires long release
RMS/Peak is a rarely used parameter, which determines the characteristics of
the release and compression behavior. RMS oriented compression is advantageous in the mastering process in order to achieve the loudest possible results,
whereas a peak-oriented compression is a pure “top-end” compression with little
effect on the overall loudness. The only native devices to provide this parameter
are the compressor from the Steinberg-Surround edition and MD3 software by
Electro/Opto: The electro mode is another description of RMS-oriented release
behavior and is modeled on analog compressors. With this release behavior, the
return speed of the VCA to the original setting increases, the closer it gets to the
original setting 0. Opto works the other way around; here the return speed decreases in line with an approach to 0. The Opto behavior is well-suited for drums,
for example. Some Waves plug-ins offers the choice between the two modes.
The crossing at the threshold point is referred to as a knee, or more exactly, as
a soft-knee or a hard knee. For soft-knee settings, the level reduction is applied
moderately even before the threshold value is reached in order to achieve softer
dynamics. Soft-knee settings are particularly good for subtle compression. The
use of soft-knee settings is incompatible with the use of the attack parameter in
order to generate snappiness because they pursue contradictory targets.
complex routings and does not permit the synchronized editing of stereo tracks.
The new VST 3-Standard now allows side-chains; it can therefore be expected
that this function will soon be used in many VST plug-ins. Nuendo 4 and Cubase
4 now has VST3 plug-ins providing side-chaining flexibility already familiar in
the analog studio world.
Gain reduction – or “GR” for short – generally refers to metering and shows
level reduction on a display. The greater the displayed level reduction, the greater
the compression. The speed with which the display jumps back to zero indicates
the release time setting. If the ratio is set to 1:1, nothing is displayed since there is
no compression. If the compressor that you use has no automatic make-up-gain
function – in other words, if it does not increase the output level in line with the
compression – you can use the gain reduction display as a guideline for raising
the output level. Once you have cut 6 dB from the peaks, you can raise the output level by 6 dB. Remember, though, that an honest A/B comparison of your
compressor settings is only possible if you listen to the original and the processed
version at the same volume setting.
The gain reduction display is useful for visually monitoring the threshold, ratio,
release, and output gain parameters.
Compressor Types
Single-band Compressors
Compressor with a soft-knee setting
The term Side Chain is borrowed from hardware devices and either refers to a
separate physical input (side chain input) or a signal path captured from the input
signal for peak detection. This independent signal can generally be processed
outside of the monitored signal with an EQ, generally a double bandpass. For
example, you can remove bass parts from the side-chain path to prevent the snare
compressor from reacting to every beat of the bass drum. In a practical studio setting, a discrete physical side chain input can be used to add a spoken voice track
signal in order to control the background music with the level of this track. The
music becomes quieter as soon as the spoken voice exceeds the preset threshold
value. This is very often used in radio, for example. On the software side of things,
the TC Native Bundle with Side chain plug-in has this feature, which is no longer available. Certain Waves stereo compressors provide the option of using one
channel as a side-chain channel and the other for processing. This requires rather
Single-band compressors process the entire spectrum as a single band. Traditional
analog devices are always single-band. This type offers all compression modes
relevant to mixing. Single-band compressors tend to produce pumping effects
more than multi-band compressors.
Multiband Compressors
Using bandpass filters, multi-band compressors split the spectrum into 3 to 10
bands – similar to a graphic EQ – in order to process them separately. The threshold of the bass area can be determined by the bass drum beat and the threshold
of the mid range by that of the snare drum, and so on. This disassociation allows
stronger compression of individual bands and avoids pumping which would result in interaction between the various frequencies. The multi-band compressor
is best used as a mastering tool. I advise against using it in the mixing process in
order to avoid over compression and also to improve the chances for later rectification during mastering. High-quality multi-band compressors for the purpose
of loudness optimization work with soft crossover filters whose steepness does
not exceed 6 dB per octave. This avoids any phase errors caused by steep bandpass filters. A further feature of a good multi-band compressor is its ability to
increase density – i.e. loudness – without pumping or sounding artificial. Here,
too, the shape of the release curve is crucial: RMS (loudness-oriented) compression has release behavior that is analogous to compressors based on Electro-controlled VCAs. The optional Powercore MD3 compressor is the only multi-band
compressor that I know of which offers different RMS and peak-oriented characteristics. The use of multi-band compressors requires precise settings and harmonization of many parameters; basically, it only makes sense when this broad
palette is used for creatively structuring the sound and not just for compression.
The different processing and leveling of the individual bands has a great deal of
influence on the overall spectral design, comparable to an EQ. If you do not wish
to process the sound spectrum, it is better to use a loudness maximizer, which is
both easier and quicker, provided that all you want is to optimize the loudness.
Loudness Maximizer
The loudness maximizer plug-in from the Steinberg Mastering Edition has given
its name to a whole group of compressors specialized in maximizing loudness,
particularly during the mastering process. Among them are also the L2 and L3
plug-ins from Waves, the Sonnox Oxford Inflator (PowerCore or native) and the
UAD-1 Precision Maximizer. Like multi-band compressors, these plug-ins are
not suitable for mixing. An exceptional case where they may be used would be
for increasing the assertiveness of a drum group. If you do not have a brick wall
limiter, the loudness maximizers can also be used as reliable limiters as long as
you make sure not to be too aggressive.
Exercises with Compressors
If you are using Cubase or Nuendo, open “Compression_Exercises.cpr” or
“Compression_Exercises.npr” in your host application. If you are using Logic or
other software, please import the OMF-2 file “Compression Exercises / OMF2.0”
into an empty project.
The unprocessed wave files 01 to 07 are on the light green tracks 1 (mono) and
2 (stereo). Insert either the recommended compressor or the best compressor
you have available for working through the exercises.
(Generally, compressors are to be inserted and only in special cases used as
send effects!)
It is up to you if you want to mute the playback tracks or not.
For comparison, the yellow-colored tracks 3 and 4 (mono and stereo) have already been processed. Naturally, they are muted at first. In some cases I have
EQ’d the examples before compression. Interfering signals that become amplified
through compression should be removed with an EQ. This includes low-frequency interference and crosstalk. If you are using the UAD-1 1176 or the Sonnox
Oxford Dynamics compressors, the presets I use are already in the fxp folder. For
this you should use the Load Effect function within the plug-in. Here you will
also find the EQ presets, if EQ has been used prior to compression. The names
FX1, FX2, etc. represent the order of the plug-ins used.
There is no surround track on the OMF version. Since the examples come from
a surround mix, I have included the surround playbacks just for fun.
Before we can start with the actual exercises, there is one final quirk: because the
1176 compressor is a true copy of the Urei compressor, the original’s somewhat
eccentric features have also entered the world of plug-ins. Among them is the
illogical operation of attack and release; the further you turn the knobs to the
right, the shorter the values. For long attack and release times, you have to select
settings between the left limit and 12:00.
Exercise 1: Lengthening a Short Bass Sustain
Use a long release time here! Choose a ratio of between 4:1 and 8:1, according
to your personal preference. The greater the compression, the greater the scope
of the amplifier when returning to its original value. A higher ratio and a higher
negative threshold therefore give you more level headroom in achieving your target: a bass that can be heard continuously at a consistent loudness. You can choose
between short (discrete) and long (to emphasize transients) attack times. Longer
attack times increase rhythmic localization. Set the display on the 1176 to GR for
showing gain reduction. Adjust the threshold value by changing the input level.
The higher the input, the greater the compression. When the display is in GR
mode, notice how much the dynamics are reduced and how fast the amplifier
springs back. The longest release time on the 1176 compressor is 1100 ms; you
will certainly find the right setting with the UAD-1 1176 or a good native clone.
However, in this exercise I prefer the Oxford Dynamics compressor for a thicker,
almost artificial result. You can even see the significantly weaker level reduction
of an individual sound. You can also recognize the transients generated by the attack time. In pop music productions, I often notice extremely compressed tracks
like this. (See page 135 with illustrations of the two wave forms before and after
Exercise 2:
Lengthening a Short Acoustic Guitar Sustain
Similar to exercise 1: I dropped in a lowcut filter in front of the compressor to
eliminate low frequency interference.
Exercise 7: Make a Funky Guitar Sound Crispier
Here we are also using Attack to create artificial transients, which refresh the signal. At the same time, the fast release time is used to increase the presence of the
quieter ghost notes, or strumming sounds, to make them clearly audible.
I have placed a lowcut in front of the compressor in order to avoid bringing
out interfering low frequency signals, which frees the sound of any disturbing
Compression on individual tracks in the mix is mainly used for shaping the
sound and controlling natural dynamic jumps. At this point, it is not used to create a thicker sound, as is the case during the mastering phase. Compressor usage
falls into three different categories:
u Supporting sustain (presence through level consistency).
Exercise 3: Louder Snare Strainer
u Supporting transients (presence through snappy artificial transients).
To bring out quiet parts, set the release time to a very short value. Use this to
bring out the strainer part of the snare sound.
u Increasing the level of quieter sections.
Exercise 4:
Snappier and More Prominent Snare Attacks
Here you will need long attack times. For the 1176, adjust the attack knob to
somewhere between 9 and 11 o’clock. The longest attack time of the 1176 is 800
Exercise 5: Bringing out Tom Attacks
Just as in exercise 4.
Exercise 6: Putting “Bite” into Vocals
You can turn the ratio up to 12:1 of this (with conventional compressors, the values will generally be somewhat less than on the 1176). The “bite” is generated by
a long attack time. Release time should correspond to the song tempo in order to
prevent pumping on the vocals. I have also used EQs here. In the mix (with full
playback) I copied (doubled) the lead vocal track and treated it slightly different
(de-correlation) and then mixed both tracks into the center. That way, the vocals
come through more clearly.
Indirectly, this processing results in restricting the signal dynamics; the compressed signals can therefore be louder and more even in terms of their respective levels.
The use of compressors in groups often requires a more subtle and inaudible setting, with low ratios where the goal is to join individual signals and achieve a subtle
restriction of dynamics. In individual cases, a loudness maximizer may be used in
the compression of groups as long as this results in the desired punch. Transient
modulation – first introduced by the SPL Transient Designer and now also available for DAWs with the same named UAD-1 version and tools like the Sonnox
Oxford Transient Modulator or the Duende Transient Shaper (drum strip) – is
particularly suitable for percussion and drum groups to inject life and spirit into
flabby-sounding recordings.
Only in the master bus should there be a limiter. This could be called a “technical
limiter,” because its purpose is simply to protect the output against overs. As a
basic rule, during the mix you should cap only the highest peaks. That is why it is
always advisable to glance at the sum limiter whenever you have raised the level
of channels or groups, just to make sure that the limiter is still only processing
occasional peaks and not intervening constantly.
If the compressor settings are satisfactory, don’t make any changes to the signal
level prior to compression, since this will indirectly change the threshold level
and therefore, the amount of compression. This could happen, for example, if
you compress in one group and then raise the level of an audio channel routed
to that particular group. Whenever you have to make such a correction, make
sure that you keep an eye – or better an ear – on the relevant group compressor.
Insert compressors either in the second or third slot on the insert rack so that
you keep your options open with regards to dropping in EQs in front of them.
In Cubase and Nuendo, the channel EQs are post, so that the channel EQs are
not suitable for removing undesirable signals upstream from the compressor. In
Nuendo 4 and Cubase 4 you now finally have the possibility of changing plug-in
order simply by drag-and-drop.
OOD REVERB PAYS OFF in the stage depth of complex mixes! But
what are the hallmarks of good reverb? In order to identify the features
of a good reverb device, we will first look at the bad ones in so many
reverb plug-ins: Unsatisfactory reverb is characterized by weak stage depth in
the final result. It may sound like reverb, but does not create spatiality within the
mix, only a greater fuzziness of sound. One of the characteristics of bad reverb
is to create the need for a lot of reverb in the mix in order to convey the spatial
information. If 99% of the presets of a reverb plug-in are things like “Big Tunnel”
and “Fantasy Cathedral” and if there are no good ambience presets, you should
be wary!
One look at acoustic reality explains why that is the case. The reverb information
of your current location is made up of an unlimited number of reverb returns that
reach your ears from all directions at various levels. Clap your hands, and you
first hear a loud peak, followed by the highly complex reverb return. However,
the reverb return is extremely quiet compared with the peak of the clap. It could
be that the reverb return is 80 to 90 dB quieter than the peak. Nevertheless, you
perceive this spatial information. This still works, even if other environmental
noises overlay the reverb echo. Our ears are veritable masters in processing and
interpreting this very quiet spatial information subconsciously, that is, psychoacoustically!
What distinguishes a reverb device from the real event? The direction of the echo
and the information density or calculation depth!
Sound Example
Fundamental changes in direction are only possible when working with surround sound. In stereo mixes, in order to avoid the masking effect, you have to
separate the reverb from the dry signal. Place the reverb signal on one side of
the panorama and the dry signal on the other. Generally, spatial localization is
negatively affected by the masking effect. Taking this phenomenon into account
in your mixing strategy is therefore a question of professional skill. Because of
the masking effect alone, the reverb part in a stereo mix has to be slightly more
pronounced than in real life, in order to convey the impression of the same spatiality.
Native and DSP-based Reverb Devices
The second aspect, calculation depth, is of even greater importance. This is the
area where reverb plug-ins struggle for a compromise between CPU savings and
good calculation depth. Allowing our ears to interpret a reverb echo as such on
this subtle level requires a certain information density. If your plug-in cannot do
that, increase the level of the reverb return in order to force the spatial information onto the ears for as long as it takes to be accepted and interpreted as spatial
information. This is almost bound to result in a fuzzy sound.
The increased calculation power of DSP-based reverb devices alone results in
their higher performance levels. The Waves APA 44, for example, “only” calculates 6 good convolution reverb devices compared with 26 channel strips. This
indicates the power hunger of serious reverb plug-ins.
Therefore you cannot recognize a good reverb device by the great sound of its
presets when stepping through them in solo mode; instead, you need to test it
to see if the reverb return still stands up in the complete mix at low levels. To do
this, take a dry drum group or a dry loop and try out your reverb devices in subtle
room settings such as “Ambience” or “Small Booth” by mixing the reverb onto
the drums and switching the reverb on and off while listening to the playback as
a whole. A good reverb does not need to be consciously audible. But you should
miss it as soon as it is switched off. Does the spatiality of the drum collapse when
you switch off the subtle reverb, giving you the impression that the drums are
stuck on a plank at speaker lever? And did you earlier perceive the drum sound
as naturally spatial? If so, then you have an excellent reverb device. (Please refer
to Sound Example #12).
On the native-based systems, convolution reverb is the first choice; unfortunately
it requires quite a bit of CPU power. Convolution reverb is calculated on the basis
of impulse responses taken from real rooms or legendary reverb devices. The
quality of convolution reverb is largely dependent on the quality of the impulse
responses. Convolution reverb cannot be manipulated as flexibly as a synthetically calculated algorithmic reverb. For that reason, it is quite popular to combine
convolution reverb with an algorithmic one.
If you do not have the CPU resources for simultaneous real-time operation of 4
first-class reverb devices, you can re-import the reverb response as an audio file
and save the preset that you used in order to implement later changes, if the need
arises. This is undoubtedly the better option rather than compromising with regards to reverb plug-in selection.
In “Overview of recommended plug-ins” you can read more about the crème de la
crème of reverb devices.
Using External Reverb Devices
If you are lacking in computing power and do not have a DSP card, a top-ofthe-range external reverb device is an excellent alternative. Since prices of digital
devices have dropped significantly, high-class range reverb devices – often no
longer available new – can be bought at a fraction of their purchase price at studio clearances or Internet auction sites. The only prerequisite for their integration is an audio interface that provides the required number of input and output
channels (in addition to the stereo bus output). Generally, these are one or two
audio interface outputs in order to send into the reverb mono or stereo, and two
additional audio interface inputs for the reverb return signal or to add it to the
mix as an “open” input channel. In your host application route one or two (for
stereo) auxiliary sends to the audio interface outputs, which are then routed into
the external device. The effect part in the reverb device is set to 100% (completely
“wet”), which is common practice for all send effects. In order to be able to gen-
erate the required number of different reverb variations for the layering of your
mix with a single device, you need to record the reverb return for different send
mixes with different reverb settings. This could be a little stereo reverb with delay
tail for the vocals, a subtle little drum booth for the drum and a large reverb just
for the snares.
Apart from calculation density, quality and room size of the selected reverb, there
are two major design concepts available with a significant psycho-acoustic effect
on the ability of our ears to interpret distance: pre-delay and frequency curve of
the reverb return. Another important aspect is the quality of “early reflections,”
depending on plug-in and preset choice.
This approach has both advantages and disadvantages: The disadvantage is the
experience that is required as well as imagination in order to select appropriate
reverb types and balance them with each other without being able to listen to
them simultaneously from the start. The advantage is the ease with which the
recorded reverb tracks can be edited retrospectively and the comparatively low
CPU requirement of audio tracks. It is easier and clearer to use EQ for managing
panorama distribution, loudness, and sound structure in audio tracks than to use
reverb return; if required, other effects such as delay can be added to the reverb.
Aspect 1: Pre-delay as Sound Design Component
With computer-based recording now the norm, the absence of good sounding
recording studios – with their rich and varied acoustics skillfully recorded with
additional ambience microphones on separate tracks – leads to more or less dry
recordings made in dead-sounding recording booths, or improvised recording
environments. This forces the mixing engineer to create artificial spaces in which
to place the sound. Otherwise a mix will sound as flat as a board. This is true for
mixes that strike us as being dry on first listening as well as highly reverberant
romantic rock styles. The difference is only that the dry mix uses discrete, small
room reverbs with first reflexions, while the romantic ballad additionally employs much more obvious spaces.
Reverb as Send Effect
Reverb is usually used as a send effect, routed through the aux buses post-fader,
in other words, the signal after the fader. The advantage of this is that the level
in the signal that is sent adjusts proportionally to any changes effected by the
channel fader. If the send was pre-fader, the effect level would remain static, even
when you mute a channel or set its level to very quiet. Because the adjustment is
done via the send level, the reverb or effect ratio in the reverb device is generally
set to 100%.
Pre-delay time is a basic and important factor for setting the distance in the reverb. This parameter is used to set the time span between the direct sound and
the onset of its first reflection. Pre-delay use is a controversial subject: a high predelay time suggests proximity of the sound event, but at the same time tends to
make the overall sound fluttery and less tight. Pre-delay times between 60 and
100 ms will already be perceived as a “slap.” If used, they should be synchronized
to the song tempo. Before I confuse you, let’s recapitulate.
Case A: You are in the middle of a huge cathedral and the sound source is a meter
(3' 3") away from you. The direct sound reaches you after 3 ms (speed of sound:
1 meter (3' 3") = approx. 3 ms), whereby the first reflections – coming from the
nearest wall 15 meters (50') away – need 90 ms to reach your ears (30 meters x
3 ms). These first reflections are followed by the actual reverb tail. Your subconscious ability to determine distances gives you an idea of your proximity to the
sound source.
Case B: The sound source is 15 meters (50') away from you and is at the end of a
cathedral and is situated one meter (3' 3") from the wall. The direct sound takes
45 ms to reach your ears, followed 6 ms later by the first reflections, which take
51 ms altogether. The direct sound, the first reflections, the reverb tail mix together, and your subconscious mind interprets this as distance. So much for the
academic explanation.
This is applicable to classical music, chamber music and small acoustic ensembles, all with carefully dosed amounts of reverb.
In my experience, this rule functions better the other way around in pop music.
I believe this may be a result of the significantly lower amount of reverb that can
be used in a pop mix, as compared to real acoustic phenomena. In this case, general theoretical doctrine cannot be applied to everyday pop music mixing.
Sounds which you want to appear direct or close by, function well with very
low pre-delay times (between 0 and 10 ms), while distance can be created with
increased pre-delay. You should be careful with placement when using high
pre-delay times with percussive sounds. All sounds relating to rhythm, such as
drums and bass, should have reverb either completely without pre-delay or with
amounts of only up to approximately 10 ms. When increasing pre-delay here,
check for rhythmic consistency.
Case A:
High pre-delay values of up to approx. 60 ms are particularly suited to chorus and
strings. This puts them spatially toward the back of the “stage.”
You should experiment with different pre-delay values and test the impact of the
results. It is best to do this with a partner, since we tend to hear what we think we
should hear.
Here are some points to take into consideration when dealing with pre-delay:
u With very acoustic, natural mixes, follow the natural behavior of pre-delay;
longer delay times for nearby events and shorter values for far away ones.
In a natural listening environment a long pre-delay
indicates a close sound source
u In pop music, use the opposite approach; short delay times for nearby events
and longer values for far away ones.
u With percussive sounds and long pre-delay times, sync delay times to the song
tempo; avoid “slap delays.”
Case B:
u If the reverb signal muddies up the dry signal, try a higher pre-delay value, also
synched to the song tempo
Since we are focusing exclusively on parameters of importance for the design of
stage depth in this section, we will also briefly look at EQing the reverb signal,
which also plays a crucial design role.
Aspect 2: EQing in Reverb Return as a Design
First of all, it’s important to remember that deep sounds have more energy than
high sounds.
Secondly, sound moving through a gas-filled – in other words, air-filled – room
loses energy with increasing distance from the source, because the air acts as resistance to the moving sound. If this were not the case, we would be able to hear
In a natural listening environment a short pre-delay
indicates a distant sound source.
all sounds, except those which pass over our heads due to the curvature of the
On the basis of these two laws of physics, it is easy to deduce that high frequencies
lose more level over the same distance than low ones. The frequency curve for
the decrease in natural reverb is therefore not linear, because the high frequencies decrease faster than the low frequencies. The greater the distance between
the listener and the sound event, the lower the proportion of high frequencies
in the reverb signal. This is why treble roll-off of the reverb signal is one of the
most effective psycho-acoustic means of representing distance to a sound source,
since our ears interpret this information subconsciously. Reducing the trebles in
the reverb of the lead vocals – which you want to place right in front of your mix
– gives the listener a contradictory message. On the other hand, you will also run
into trouble if you place the chorus in a large room, select a pre-delay of 30 ms,
and make the reverb very brilliant. Here, too, listeners are given contradictory
data and therefore have no clear spatial information to interpret. This is why your
reverb settings should have more treble for sounds that are in the front of the mix,
and less for those placed in the back.
u Rich treble content in the reverb indicates nearness, lack of treble communicates distance.
u The main ambience – usually drum ambience – should be discretely mixed in
all important tracks to create acoustic cohesion.
Basic Delay Rules:
u Delays that are intended to generate space should be tempo-synched even if
they are short (quarter, eighth, sixteenth, etc.; dotted and ternary versions).
Make notes concerning the appropriate delay times in relation to note values
for every song. With ternary rhythms, calculate the delay times for triplet quarter notes, eighth notes, and so on. Divide 60,000 milliseconds (one minute) by
the song tempo (quarter notes per minute) and you will have the number of
milliseconds per quarter note.
Here are some basic rules to help guide you. Keep in mind, however, that since
mixing music is such a varied and lively process, exceptions prove the rule.
u Slight variations in delay time can drive (shorter delay times) or drag (longer
delay times) the rhythmic feeling of the song.
Basic Reverb Rules:
u Beware of phase cancellation (comb filter effect) with delay times of under 10
u When using more than one reverb device, organize the programs or presets by
room size (small, medium, large, etc.)
u Reverb tends to blur the mix more than delay.
u The balance between space and distance can be controlled with the effect
u Reverb length – particularly with gated reverbs and snare reverbs – should be
tempo-synched. Snare reverb usually ends on the next full beat.
u Reverb with very short decay times create discrete spaces.
u The longer the decay time, the earlier (with regard to effect level) distance is
created by the effect.
u Lead vocals and reverb: it's all about the mix! Where a lot of reverb is needed
(for example with ballads), create a dry counterweight by doubling the lead
vocal track; de-correlate the track by adding EQ, compression, and possibly a
short delay and mix the dry lead track back in. This way, the lead vocals don't
get pushed to the back too far, but at the same time sound fatter.
u Single delays between 10 and 30 ms long can discretely "thicken up" the sound,
while the localization of the first sound event remains intact independent of
where the delay comes from. The human ear perceives single delays between
10 and 30 ms as direct sound events.
u Delays up to 30 ms – especially stereo delays – are well suited for making the
sound richer with a slight room effect.
u Delays between 30 and 60 ms produce what is called the doubling effect
(Beatles sound).
u Delays between 60 and 100 ms are called slap or slapback echo (Elvis sound).
u Stereo delays of up to about 100 ms create the impression of an acoustic
u A delay time above 100 ms is usually referred to as echo and tends to suggest
distance and space.
u The longer the delay time, the more the sound appears to be indirect.
u Delay tends to blur sound less than reverb.
u To discretely create space, delay should be very subtly used, so that you miss
the muted FX channel, but you don't really consciously perceive the delay.
u Echo (delay longer than 100 ms) which is not tempo synched is good for creating an effect that is clearly heard as such in the mix.
So that all of this does not remain dry theory, open a project with vocals and try
out various settings.
Pre-fader or Post-fader?
Why is the reverb effect used as a send effect? There are several reasons for this: it
is the only way for you to have the option of sending several sources to one effect
device; the only way to edit the effect-return signal separately with panorama,
EQ, or other effects, as may be required, without affecting the placement of the
dry signal in any way. In rare individual cases it may make sense to insert a reverb
device. To ensure that the send works properly, make sure that the reverb device
is set to 100% effect – often referred to as “wet” – and the aux-send to “Postfader.” “Post” and “pre” refer to where the signal is tapped on the channel signal
path with regards to the channel volume fader.
Generally, the default setting for aux-sends in DAWs is post – in other words, after the channel fader. The advantage is that the level of the tapped signal remains
proportionate to any change in the channel fader. If you have already determined
the right proportion of reverb, this proportion would change if your aux-send
is pre-fader and you make any changes on channel fader, since the dosage is no
longer linked to the channel level. The effect amount remains the same while the
dry signal changes.
The “pre” setting makes sense with headphone paths, so that the engineer can
create independent studio mixes without affecting the headphone signal in the
overall mix. In rare exceptions, pre-fader aux-sends for effects do make sense.
For example, when you want a sound to disappear in the reverb, set the aux-send
to pre-fader and slowly bring the channel fader down. This way, the dry signal
disappears in a soup of reverb. This effect was popular in the 1970s, but is seldom
used today. Occasionally, using reverb post-fader for quiet, legato strings that are
meant to be in the back of a large room can be problematic, because the signal
sent to the effect device is too weak, and too little reverb comes back. In this case
switch the send to “pre” and reduce the channel fader level until the relationship
is correct.
How Many Reverb Devices Do You Need?
Planning Depth Design
I know of no logical rule for this, other than that there is no rule. If the tracks already have a good, natural ambience from a recording room with a good sound,
and if the final aim is for the production of an intimate sound, it may be possible
in extreme cases to do without reverb altogether. However, if you have accepted
the challenge of a very complex mix with 80 to 100 tracks with many events that
overlap spectrally, using up to 8 reverb devices can make sense. Whatever the
scenario, at some point – apart from reaching the limits of computing capacity
– you loose the ability to differentiate and using more reverb devices does not
make any sense.
For a typical pop music setting with a five-piece band, a handful of doublings and
a bit of ornamentation, I imagine the following scenario: in line with the panning
strategy outline as developed in the section on “Panorama” regarding the placement of individual instruments, the drums are placed at the back of the stage. For
this, we use our send effect 1 with a subtle little ambience reverb or stage reverb.
This reverb creates just enough information to bring the dry-recorded drums (or
carefully programmed and totally dry samples) out into the spatial dimension.
In context, you will only be able to identify the reverb as such if you listen very
carefully. Only by switching it off again will you notice the absence of spatiality.
To make sure that the bass does not sound like it’s coming from another room,
it is only given enough of this reverb to appear as a unit with the drums. Please
remember the advice given at the beginning of this book: use the channel send
paths for controlling the effects in order to adjust individual doses more accurately and use the group track/solo function for dry solo listening. I like to give
the bass drum a little less ambience than the other tracks to keep from having too
much ambience from the bass drum skin sound (upper mid-range). Otherwise,
the bass drum can sound fluttery. If the mix does not sound cohesive, then it may
help to give the other important instruments some of this basic ambient reverb.
This creates a feeling of unity.
Depending on style and genre, an additional, more obvious reverb is often used
for the snares, giving them more “space.” Be careful, though, because you might
find yourself on a trip back to the 1980s. For ballads, we can also apply the send2-reverb to the toms, making it more impressive. If necessary, reverb 2 can be
gated by inserting a noise gate into the reverb return path and using the gate to
reduce the decay length. This gives you the option of using a full, long reverb that
is artificially shortened. This trick also originates in the 1980s and must be applied with caution and sensitivity.
The gate – or decay – time should be synched to the song tempo.
If the snare you want reverb on is within a loop or is recorded live (un-gated) with
crosstalk from other drum tracks, then you should not use reverb on the whole
track. To avoid sinking the track into a muddy reverb soup, copy the track and
specifically gate the signal to send a short, punchy impulse to the bigger send-2reverb. If, after creating consistent panorama distribution, you want a modern,
relatively dry sound style, you can pan the guitars and keyboards without reverb
right and left. This way, they sound like they are very much in the front. A delay
is sometimes more helpful for a “mud-free” and differentiated creation of space.
The sounds that are treated this way will stay up front. If backing vocals, percussion, or other instruments are placed consciously toward the rear, we will need a
large reverb with some pre- delay and filtered trebles. Reverb can be generously
applied here so that the spatial information is perceived by our ears, since the
masking effect tends to minimize the perceived reverb amount. Try using a stereo expander in the reverb return. If the result is unconvincing or not mono compatible, try two identical reverb presets from two different devices, panned left
and right. Both reverb devices receive opposite send signals so that the cabassa in
the left of the panorama is reverbed to the right and vice versa.
sized reverb space. Combinations of reverb in delay or delay in reverb can also be
used to this end (see below). Or combining a discrete early reflection reverb with
a delay sent to a spacious room reverb can make the vocals sound bigger. If you
find good settings for this kind of work, I recommend creating your own effects
channel presets, so that you have quick access to your reverb setups.
In the following scenario, 4 to 5 reverb devices are all that you need, provided
that for the large reverb space, you just need one reverb device with a stereo LeftRight return. It goes almost without saying that send effects 1 and 2 have no, or
only a very short pre-delay. Send effect 1 can have a slight treble roll-off to define
the sound’s position, whereas the snare reverb can be without EQ (trebles not
filtered) in order to place the snare at the front.
Here is a quick summary for the setup used in this example:
Send effect
Varieties of reverb/FX
Area of application
Treble roll-off
Send FX 1:
small reverb/ ambience
drums and some bass
0 to little
Send FX 2:
medium–large reverb space
Send FX 3:
big room
background events
(chorus, strings)
up to 60 ms
Send FX 4:
lead vocals
0 or 1/8th note no
Send FX 5:
on style
delay or reverb/delay
lead vocals (if nec.)
Send FX 6:
for ex:
(decent) tempo-synched delay
guitar & keyboard (if nec.) L10/R20
Send FX 7:
delay (strong)
Send FX 8:
The best comes last - reverb for the lead vocals! Especially in mixes that tend to be
dry, vocals require a particularly high-quality device to prevent them from being
pulled into a cloud of reverberation. Sometimes all you need is a subtle, small,
and unobtrusive reverb with attributes similar to a drum booth. Often combining with an additional delay works well, which might blur less than a medium-
If you decide to go for the option of a left/right split reverb return for the large
reverb, you need an additional device. I cannot give you any detailed information
regarding delay parameter settings – this depends too much on the individual
song and style. Keep in mind that instruments placed laterally in the panorama
will be pulled towards the center if their effect returns have centered pan settings. This phenomenon increases with effect amount and is particularly noticeable with modulation effects. You can counteract this with extreme panning on
the dry tracks. If this is not possible, use a second panorama-weighted effect, or
insert the effect into the track instead of using it as a send-effect.
Since music styles and tastes are both individual and varied, please use the example here as an inspiration for employing your own selected effects, which
should serve to support your layering strategy in accordance with your outline
(see “Dimension 1”). Make notes on the applied effects – like “Send FX4 – small
ambience; only for lead vocals” – on a separate sheet of paper. A number of DAW
programs let you make notes on each track, which might help you to avoid having too many paper notes lying around.
The time and effort of using left/right-split effects is only justified if the reverb
part becomes too large to convey all the spatial information as a result of the
masking effect. The more complex a mix, the more time and effort is often required for placing all events accurately within the three dimensions.
Delay Instead of Reverb
A delay tail has the advantage that it can make lead vocals sound fuller, giving them
more volume without endangering their frontal placement. The more obvious the
delay appears in the mix, the more attention you need to pay to the stereo width of
the effect in order to ensure that the lead vocals retain their center position.
In conjunction with delays, there are two interesting variants of ducking, or suppression. Either the tail only appears when the lead vocals are silent – in other
words, there is no input signal over a certain period – or the tail is suppressed as
soon as there is no signal. The first option is useful if the delay blurs pronunciation clarity too much or if it is stylistically undesirable,
but does however create a good filler effect during
pauses. The second option is useful for fattening the
vocals without being too obvious in the vocal pauses.
The T.C. Electronic TC2290 Dynamic Delay
On the hardware side, the TC 2290-Dynamic Delay is
well known for this very purpose. Unfortunately it is no longer produced, and I
am not yet aware if it is available as a virtual plug-in. However, this is not a big
problem for computer-internal mixing because the effect is easy to reproduce by
exporting the delay as a single track (mono or stereo) and then – using the “import into audio track” function – linked back into the mix as an audio track. Then
it is very easy to edit the desired effects manually using cut, fade, mute, and level
tools. Working with computer keyboard shortcuts is conducive to speedy work.
While delay also creates a spatial impression, it is not suited for accurately creating
specific stage depth for different sound events. This is why delay should only be
used to as a stylistic aid. In some British styles, delay plays a larger role in layering.
Dub delays, for example, as used in dub and reggae, give the impression of a certain distance. The device customarily used for
this is the Roland RE 201 Space Echo. The combination of tape
echo with spring reverb gives this effect its unique character.
Every good dub band has an extra member just to operate this
device. Because for technical reasons, this tape-based device is
quite stingy in the higher frequency domain, it communicates The Roland RE 201 Space Echo
distance and spatial depth when heavily used. An authentic reproduction of the
Space Echo is available as an optional plug-in for the UAD-1 card.
Delay in Reverb or Reverb in Delay
Ever since it has been possible to make complex effect cascades with most DAWs,
a whole field of possibilities has been opened which has always been part of the
standard repertoire of traditional mixing consoles. The effects return can be routed back to a full channel and, using aux-sends, be routed to any other effect. In
a virtual mixing console, this is not quite as straightforward, because the host
application would have to constantly maintain correct latency time correction
throughout the signal path; so far, this has not yet been achieved on such a complex level.
In order to use effect cascades in spite of this limitation, some host applications
make it possible to cascade effects – one after the other – on a single channel. In
Nuendo or Cubase, so-called effect channels can be opened which let you plug
the main effect into the first insert slot. Effect channels are full-fledged channel
strips, which, for reasons stated above, have no send-target choices available in
the aux-sends, but serve as effect-return channels. The seven remaining insert
slots in this channel can be used for wide-open sound design possibilities.
I recommend trying different variations for yourself and to develop your own
presets for various set-up uses. This is very easy to do by using the option to save
complete channel settings. If your DAW does not allow effect cascades, then export and re-import the effect return as an audio file. Effects can then be inserted
or the signal can be sent to any desired effect. The classic combination is a reverbed delay or a delayed reverb. Other combinations with chorus, flanger, compressor, gate, and stereo expander effects are also possible.
On the accompanying DVD you will find a special “Layering Exercise” as a practical application of the theoretical information on stage depth we have been discussing.
This exercise is provided as a .cpr file for Cubase, an .npr for Nuendo, and OMF
for all other programs. You will find it in the “Exercises” folder.
In the .cpr and .npr files, the effects used have been left in place. If you get a message saying that one or another plug-in is not available on your system, save the
message in a text file and replace the plug-ins with preferred plug-ins available on
your system.
been copied and the snare manually shortened, in order to create a short send
signal for the somewhat larger snare reverb. Here you can experiment with a
thick gated reverb.
Send FX 3 is a normal tempo-synched delay, providing some movement on
tracks 11 (clave), 12 (triangle), and 15 (Lounge Lizard arpeggio) and places both
events slightly toward the back as a result of the space that has been created.
Send FX 4 gives the triangle and the Lounge Lizard e-piano on track 14 a discrete
spacey sound with a filter-modulated OhmBoyz delay.
Send FX 5 finally addresses the issue of reverb layering. This send gives the percussion tracks a medium, thick room from the Waves IR1 convolution reverb
plug-in. The return is processed with a little stereo basis widening in order to
counteract the masking effect and to consciously place the percussion behind
the drums. I have put a little pre-delay on the reverb and slightly attenuated the
trebles. The bongos are consciously far to the left and the conga fills far to the
right in order to remain unmasked by the other sounds.
Send FX 6 gives the chorus a warm room with the UAD-1 Plate 140 Reverb and
the Send FX 7 provides the lead vocals with the impulse-response based Vocal
Chamber from the Wizoo W2 Reverb.
The aim of the exercise is to demonstrate how various kinds of reverb provide
different results in the mix. Try different devices and compare the results. Make
sure that the reverb devices and presets you use deliver the desired spatial information in the context of an entire mix.
The tracks “Mixdown wet” and “Mixdown dry” have been created for reasons of
comparison. Naturally, the amount of reverb – like so much in the mix – is a question of personal taste; other results are not necessarily better or worse. Have fun
setting this up!
The Send FX 1 is being used for a medium-sized drum chamber. The goal here
is to create an unobtrusive space for the drums. You may want to use an SIR2
convolution reverb or another convolution reverb device, such as Altiverb, for example. The sends of channels 1 through 5 (overhead, bass drum 2, loop, fretless,
and snare) go to a medium-sized room. Use the solo button on the folder track to
hear “wet” drums and the group track solo button to hear them ‘’dry.’’ This works
because the send level is taken from the audio tracks and not the drum group
If you are using OMF: Tracks 6, 13, and 17 are group tracks in the Nuendo and
Cubase projects and therefore have no track numbering in your OMF project.
(Bass drum 2 on track 2 is slightly gated in order to make it crispier and the loop
track 3 has a little bass reinforcement at 100 Hz. The drum and percussion group
tracks have a little compression).
Send FX 2 is only for reverb on the already shortened snare track 5. The loop has
OME PRODUCERS TEND TO RECORD more in a recording session
than can be intelligently used in the mix. At the beginning of a mixing
session, it is not uncommon for many instrument tracks to be present
from start to finish. The mixing process can then be used to make a selection for a
dynamic and interesting song structure. Few things are more boring than a song
that starts with full instrumentation from start to finish. The mute switch is one of
the most effective tools for accomplishing a good, dramatic song structure.
A division of labor between recording engineer and mixing engineer ensures
the required dose of impartiality needed to experiment enthusiastically with
muting some tracks in different parts of the song. This also opens new design
options for the creation of entirely new material, such as vocal or groove parts
(Sound Example #01). Interrupting loops, shaker figures, etc. can improve the
song. For genres like techno and hip-hop, these methods are even part of the style
techniques. Numerous variations are first created on top of a loop and are then
arranged into the mix or simply eliminated. This method involves the creative
aspects of arranging deep in the mix process. Fear of the mute button can quickly
result in overloaded and unclear mixes. You must emotionally remove yourself
from the painstakingly recorded and edited tracks in order to work towards the
best possible result, and not be influenced by egos which might be affected by
using the mute switch.
Sound Example
In the mixing process, the mix engineer and producer digs into the large quantity
of tracks, which are in individual parts or individual positions unmuted. Back in
the days of analog mixing, the engineer or producer danced back and forth in
front of the mixing console, hammering mute buttons, and pulling faders up just
at the right moment. Today we have it easier even though the athletic aspects of
the activity are unfortunately replaced by the computer mouse. Regardless of the
DAW we are using, we have three possibilities:
therefore not clearly perceptible. Why should an event be allowed to create muddiness and not be readily recognizable on your studio monitors, even though you
know that the event actually exists? Every time you mute an event that does not
have its place in the mix, your overall sound will become clearer.
u Clip-based muting of particular events
If you want to keep the individual events, look for other places in the song where
there is enough room for them. That is how your mix becomes transparent and
u Mute switch (in automation mode)
u Fader automation
For the sake of a logical workflow, I suggest starting with clip-based muting. In
the section “Automation,” we will explore this method more deeply. I can already
tell you one thing; I strongly suggest avoiding automation until the static mix is
finished. The static mix represents the best possible result without resorting to
automation. Personally, I do not believe in mute automation – this feature comes
from the days when muting individual events in an arrangement window was not
possible. Avoiding unnecessary automation prevents potential irritation through
accidental deactivation of automation or moving the parts in the arrangement
independently of the automation data.
Fader automation should be reserved for detail work, rather than bringing tracks
in and out. We will discuss this further in the chapter on automation.
The question raised is such; does it make more sense to “turn all tracks on” and
then reduce with the mute switch or is it better to develop the mix by gradually
unmuting tracks and events?
Both methods are possible! However, you should decide for one or the other, depending on the style and the situation. If you use the first method, then try to find
a valid place (within the three dimensions) in the mix for every event. Once you
have found a place for all events and mixed them, reduce! You especially reach for
the mute switch if you have come to a point where it is no longer possible for all
events to transparently coexist. This way of working is often used with electronic-oriented styles, while the second approach is better for a pop song. With the
latter, start first with the framework – the rhythm group, then the warmth-giving harmony instruments, then the lead vocals, and finally any ornamentation.
Pauses in the vocal tracks can often be filled with less important events like guitar
riffs or percussion figures.
For transparency in a mix, you should radically mute all events for which you
have found no clearly-defined place within the three dimensions and which are
If the song does not have an end, you can sometimes creatively mix interesting
endings into the arrangement. If the producer doesn‘t approve, go for the good old
Speaking of fade-out! Please, never automate or process fade-outs into your
tracks! Fades are only to be done after dynamic and EQ processing has been dealt
with in the mastering phase! The only exception is arranged fade-outs, which
cannot be done on the finished master. These are fade-outs where individual
tracks are faded out against other tracks, or where the individual track fade-outs
are staggered with respect to one another.
What is the opposite of mute?
Courage! Even though courage is needed for muting, more courage is necessary for the opposite – conscious emphasis. Have the courage to bring individual
events such as effects, percussion instruments, and sound design elements briefly
to the extreme front of your mix. That could turn your song into a hit because
something out of the ordinary grabs the attention of your listeners. The strategy of first creating a static mix is one thing. But the last 10 percent of the mix
involves the really hard work and at the same time the evolution from a good
rough mix to a release-ready end result. Don‘t be misled by the results of the static
mix – maybe you need to play around with some percussion parts, for example;
there could be three clave beats placed in a pause right before the refrain with a
tenfold level increase, giving your song the decisive kick. Or a bongo lick during
a vocals pause right after the bridge. What about the screwy guitar sound in the
intro? Couldn’t it be rearranged and placed right up front to make it another song
hook? How many seconds does it take you to recognize a hit that you have heard
only twice on the radio? Go for the extreme: modify your sound with courageous
changes in individual spots. Keep your ears open to sound events in the mix that
have something special, and bring these sounds out. Act like a stage actor who
exaggerates gestures so that the people in the very last row will get the message.
Consciously listen for such stylistic techniques on your favorite CD. What makes
a sound engineer successful? Hundreds of hit song mixes demonstrate the skillful
feel that professional engineers have for bringing out that particular thing in the
song, emphasizing its charm and character, and turning a good mix into a hit mix.
Sound Example
Filter plug-ins are
suitable for creating
interesting effects. Here
is the “Filtroid” plug-in
from PowerCore.
Now that you have exhausted all of the structuring possibilities with muting or
event exaggeration, we still have the infinitely deep toy box called “Special FX” to
make our mix more interesting. The possibilities are so limitless, that it is difficult
to sum them up in words and to logically present them. Use your playfulness and
inventiveness and process the available audio material in your mix by chaining all
imaginable plug-ins together.
The repertoire includes the extreme use of all known effects, deforming sounds
through uncommon chains, reversed sound, using synth filters, vocoders, and
much more (Sound Example #02).
Using guitar amp simulation such as
NI’s Guitar Rig or Leslie cabinets are
also popular techniques.
Extensive filter effects are possible with
the “Filtroid” plug-in from PowerCore
(illustrated). Along with a reversed audio file, you might create an interesting
song intro. Make a catchy intro with
recognizable elements, preferably from
the refrain. There are no limits to experimentation! The lack of large recording
rooms has also resulted in increasing
use of a classic method; re-amping. This
involves playing a track into an amp
through re-recording in a microphone
booth. Play the track through guitar
amps, bass amps, or portable radios and
re-record the result. Or you can play a track through monitors in an interesting
acoustic space and experiment with capturing the room acoustics with a boundary layer microphone. Maybe use a pair of headphones as microphones and at-
tach them to an empty bottle sitting on a bass box. And so on. Take your time
looking for the right intro for a track that you hope will get airplay. Many legendary sounds were created accidentally. Be open to acoustic accidents in the studio
and cultivate them. Maybe you‘ll create a new cult sound!
Playing with space is certainly part of the special FX bag of tricks. Make some
special events of your mix occur in an other room (where it makes sense with
regards to content).
For inspiration, listen to Trevor Horn’s mixes. On the Seal albums you can clearly
hear how the vocals room changes for just a measure or a few words. This draws
attention and makes listening to the song interesting, even after the fiftieth time.
Two Sound Examples are included (#03 and #04).
Since creating good stage depth requires good reverb plug-ins, you can export
the reverb return into an audio file to save computing capacity. That way you can
have two to four real-time reverb plug-ins for the basic layering of your project.
Sound Examples
#03 & #04
The optional and relatively expensive VSS3 Reverb from PowerCore is a real
The VSS3 has particularly high-quality early reflections. The screenshot shows that the
actual reverb tail (Rev. Level) is off and the original (Dry Level) mixed only with the early
reflections (Early Level). In this case, the device is used as an insert.
Early reflections
are very
important for
find. It is one of the rare reverb devices that boasts excellent quality for the early reflections, which can be totally separated from the
actual reverb tail.
Early reflections are very important for representing space and many
devices do not do a very good job of simulating them. The VSS3 creates good, discrete effects that can subtly breathe new life into otherwise boring recordings.
n case you don’t create a hit on your first try, don’t worry. It is certainly
because of the arrangement. No joke! A well-arranged composition
is easier to mix than a badly-arranged B-Side track. If you are not under
time limits, stand back and try to achieve some perspective. This will give
you a more fresh and unbiased approach to your work.
NLY FIFTEEN YEARS AGO a studio owner or engineer would
have felt lucky to have motorless fader automation with just mute
and VCA automation.
Motorized faders were lovingly called flying faders, which – as opposed to VCA
automation – equipped with servos would follow the actual fader position. These
did not have all that much in common with motorized faders of modern digital
mixing consoles or DAW-controllers that serve only to enter and visualize the
electronically processed digital values.
Today – keeping with our theme of internal mixing – DAWs make dynamic automation of practically every imaginable event possible. However wonderful this
might be, the other side of the coin is not just the lower caloric expenditure of
active mixing engineers. A mix without automation on a console that is several
meters long could be a sporting event that today is reduced to training the index
A clear work strategy is the prerequisite for taking full advantage of automation.
We will examine automation strategies; automating too early inevitably creates
more work than necessary and obstructs the workflow.
This is why I believe it to be urgent to first create a static mix that reaches a 90%
satisfaction level, especially in terms of panning. As the further development of
the rough mix, the static mix is both a waypoint of departure and a guide for any
automation that is to be added.
If you start with automation before all the basic relationships are adjusted, you
will most likely have to redo most of it. That is only fun with a touch-sensitive
motorized fader, which begins writing data when you touch it and stops when
you let go. But it can be nerve-wracking when you begin with automation before
the arrangement is even finished. If this is the case, make sure that the automa-
tion data is always fully linked to the respective audio data whenever you reposition or copy elements in the arrangement.
If any tracks or parts of the song appear too loud or too quiet while you are working
toward the static 90% mix (before automation), I suggest you make the changes
“offline.’’ For example, if the lead vocals are only too quiet in the second verse, you
can select the audio events and make them louder by a specified amount. Every
DAW has this capability. In Nuendo and Cubase, you can use the Event Volume
function for increasing or decreasing levels (click and drag
the blue square on the upper
edge of the event). For precise
changes of individual beats or
syllables, cut the events with
the scissors tool so that you can
tune only the particular event
in question.
For programs that do not have
this particular function, you
can at least use an offline function (usually called “Gain”)
for changing levels of selected
audio events. In Cubase and
Events in the verse have been selected and their levels have been changed.
Nuendo, the “Gain” function
is in the submenu Process, which is under the Audio menu. Make sure that the
changes only affect the selected passages or events.
Every DAW has its own way of handling audio data; being familiar with this is
a necessary prerequisite for good, smooth working methods. Use your DAW’s
First obtain level
consistency on the
individual tracks,
and then use a
compressor. Use
automation only
after these steps.
I usually specifically deal with lead vocals in this manner, making subtle
corrections with the event volume sliders and checking each syllable for
comprehensibility. If necessary, I cut the takes into many small clips. The
compressor does the rest of the work, but you need to monitor it because
level changes that are too strong on the event input side will automatically change the amount of compression as well! The compressor’s level
amplification function generally reduces the amplitude of your manual
The audio event level can be changed by moving the blue square. The first value
shown is the absolute event volume compared to the original and the second
value is the amount being currently changed.
More complex level changes can be easily drawn in on the event level without
resorting to automation.
changes and in certain cases might even work against the result you are seeking.
This is because the compressor comes after the audio event in the signal flow.
The audio event in the arrange window of your DAW can be compared to an
analog tape machine, with its signal going into the mixing console to then be
At the very latest, when the “tape recorder level” corrections on the event level
are nullified by the compressor, it is time to employ fader automation. This interactive manipulation between event volume and compression dictates the only
logical working order.
Using this method reduces the need for automation to only a few cases as volume
automation only makes sense in certain situations:
u If compression is working against the attempt to change levels, then it is better
to use fader automation (after compression in the signal path).
u You want to create level changes that are not possible to do offline (eventbased) in your DAW.
Sound Example
Typical volume automation for
introducing a new instrument
u Your mix is 90% finished and you now want to refine the first few bars of instruments or events that occur for the first time in the arrangement. This is a
classic use of fader automation. To introduce a new instrument to the listener
in a complex mix, when a new instrument is brought into a complex mix,
the level is brought up quite a bit higher for about a measure so that the new
instrument cannot be ignored. Then it is quickly brought down to the
appropriate mix level. Since the new instrument has been consciously
perceived, the listener will be able to hear it even at quiet levels. For this
method, it is also important to keep the static level as the reference,
making just a little “bump” in the level curve at the beginning of the
newly introduced event. In the illustration showing the introduction
of a new event, you can see how this looks in the automation (Sound
Example #05).
u Programmed drums rarely have the natural dynamic of authentic
recorded drums. The verse, bridge, and chorus might be well arranged,
but sometimes the drums do not follow the dynamics of the rest very
well. In this case, level automation on the drum group track can help. In the
verse, bring the level down a little. Be careful! If the sends are coming from
each individual channel – as I recommend – then when the group level is reduced, the reverb amount is proportionately increased, since this happens prefader. The send signal remains unchanged while the dry signal level decreases.
Using parallel automation on the effect track of the reverb will help out in this
case if required.
u You want to use arranged fade-outs (staggered fade-outs of individual tracks),
which help to create an outro. Do not touch the fader on the master bus, so that
you give the mastering engineer material with full bit depth for dynamic processing.
u When using automation on tracks, remember that any further changes on
automated parameters can only be made using automation. For example, if
you made the lead vocals louder starting in the third chorus using automation, then realize that in the first verse, the track should be a little quieter; this
change should be made on the automation track or with touch-sensitive faders. Sometimes you might not notice or remember that a track already has
automation data written on to it and you might correct something using the
channel fader. As soon as the song is played again and reads the automated
value, the fader jumps back to that automation value. This is why you should
keep your eye on the green (usually) "read" button, which indicates the activated track automation.
After successfully recording all the volume automation data on to a track, you
might notice that the whole track needs to be louder or quieter. Every DAW lets
you select all the automation data for a track so that you can comfortably raise or
lower the entire set of values at one go. Before you get frustrated after accidentally
moving all of track's automation data along the timeline, find the keyboard shortcut in your DAW that locks automation data in place in the time axis.
Skillfully using panning automation is significantly more important for a transparent mix:
u Use panning automation to deploy your panorama strategy systematically,
which involves intelligently panning instruments and events to the left or right,
depending on their spectral appearance.
u Use panning automation to clear up individual spots in your song that are not
transparent. For example, if there is conflict between your lead singer and the
background vocals in only one measure, then push the less important background vocals farther out – just for this one measure to make room for the
lead. Then bring them back to where they were before. As an example, think of
a boy or girl group with four or five singers on stage: for each solo spot, the lead
singer comes a little forward and the background singers move to the side. Do
the same to make room on the pan level of your mix (Sound Example #06).
u If the above-mentioned technique does not work because the colliding events
are already widely panned, use a stereo expander. Leave it in the neutral "0" position and then – with automation – spread the stereo basis specifically on the
trouble spots (Sound Example #07). Don't forget to check for mono compatibility (using the mono button) whenever you heavily use a stereo expander.
Sound Example
Sound Example
Work through your mix project until every event is truly well defined. This may
mean panning the guitar far to the right in one refrain, while up to the bridge it
had panned all the way to the left, only because another event in the refrain is
now occurring on the left and would have collided with the guitar sound. After
all, musicians walk around the stage! Systematically deploy your panning concept. I believe that spontaneous panning jumps are always justifiable in “manufactured” music. For a jazz trio I would certainly avoid panning jumps.
Any other automation consists of fine tuning in the areas of effects and EQs. This
depends on the material you are working with, making it impossible to lay down
any set rules.
If we limit our discussion of EQ automation to one parameter, consider one that I
often automate is the cutoff frequency of a lowcut or shelving filter. On the background vocals group, try switching the cutoff frequency between 120 and 400
Hz depending on whether the group is alone and needs fullness or only needs to
back up the lead vocals. You can just as easily apply this to acoustic guitars, which
sometimes need to be full-bodied, and at other times just need to discretely play
along in the background.
Automating effect send levels is a basic part of good, clear mixes. Reduce effect
levels in airy parts of the song, when reverb would suddenly stand out. Or drop
one single word into a long reverb tail, which must be made louder for one measure in order to clearly convey the effect.
Here is an interesting concluding fact concerning fine tuning – the finishing
touches in the last 10% of a mix involving automation and small corrections usually take 80% of the total time for the mix. If the static mix takes 3 to 4 hours to
get to a decent rough mix or demo level, then you need to count another 12 hours
or a day and half is needed to polish it up for it to be releasable. If there is a lot of
preparatory work for cleaning up and reworking tracks that are delivered to you,
then you will have to reckon with even more time.
After all of this detail work is finished, I am often hardly aware of everything I
actually did and how long it all took. The large amount of time spent is dependent
on how much experimentation is necessary to achieve the result you imagined.
Since I have consciously concentrated on user knowledge in this book, I won’t
go deeper into the practical ways of using the various automation modes here.
Please refer to the user manual of your DAW to explore workflow automation
more deeply.
T THE EXPENSE OF REPEATING certain details from other parts of
the book, I believe it is important to provide you with systematically organized information and tips for working with individual instruments.
This way, you have a reference text that may be useful if you find yourself in a dead
end with your mix and you need to find some inspiration and solutions to get you
moving again.
Processing live recorded drums is a big challenge. But you can learn and get ideas
from working with programmed drums and loops. On the tutorial DVD Internal
Mixing Vol. II, the techniques I am describing will be developed step-by-step
with the help of practical examples.
Bass Drum
The bass drum definitely belongs in the center.
Often there are two tracks concerning the bass drum. One is recorded with a
microphone placed close to the head representing the “skin” sound, and then
another, large diaphragm mic on the resonance head or hole, delivering the lower
frequencies. Whether there are two tracks or just one – in both cases, the bass
drum has the same characteristic core frequencies which have their importance
and need to be treated as such; the lower frequency “bottom” range between 65
and 110 Hz and the skin sound between 3 and 5 kHz, occasionally up to 8 kHz.
For a good bass drum sound you can use a bell filter to support the lower end. Use
a medium Q-factor. For weak bass drums in the bass range, the UAD-1 Pultec
EQ can be your silver bullet. Select 60 or 100 Hz and give the boost knob a good
turn. The attenuation knob can also be used to reduce unnecessary mid-range
frequencies in the bass drum sound.
With a second bell filter, you can emphasize the skin sound to accurately control
rhythmic localization. If Q-factors are too high, the drum quickly sounds artificial. Depending on the bass drum, characteristic core frequencies can spread
through the entire upper mid-range from 2 to 8 kHz.
Now you are able to control two important characteristic core frequencies and
you now need to reduce: If the bass drum is very low, you can use a lowcut filter with a steep slope at around 30 Hz to keep the bass range clean. The lower
mid-range between 120 and 350 Hz is even more important; here the bass drum
doesn’t play an important role and you can drop in a bell filter to reduce the
sound if it is too prominent in this range. This way you free up space for the bass
and all other events that need this frequency range much more. If you have two
tracks, as mentioned above, you can use them both to shape and bring out the
characteristic core frequencies. The track with the skin sound can have a higher
lowcut in order to eliminate all low frequencies. To make dealing with such a pair
of tracks, I recommend doing one of two things. Either you can group them once
you have set the level relationships between them, so that any level changes you
make thereafter affect both. Or, you can set up subgroups. In this case, you would
have a mono bass drum group track that brings together both bass drum tracks
and then goes into the drum group.
Compressors and Gates
Compressors fill two roles with bass drums – to reduce dynamics for eliminating occasional individual overs (top-end limiting or top-end compression), or
to create more punch by artificially adding transients by increased attack times.
The longer attack times increase the percussiveness and the localizability of bass
drums (see Compression Exercises).
In software-based music production, gates have fallen somewhat out of fashion,
since you can manually separate unwanted noise from the desired sound. For a
while, the only software-based gates that existed – or that were called gates – were
not real gates. Nowadays there are true software gates, which are on par with
hardware components such as the classic Drawmer DS 201 and can be effectively used for sound shaping (for example, the Sonnox Oxford Dynamics for
PowerCore). Gates are really nothing more than the opposite of compressors. If a
certain threshold level value is not reached, then the signal level will be reduced
according to the same parameters as those found on compressors. This effect
allows you to separate the desired signal from crosstalk, but can tend to sound
artificial. With a bass drum, this aspect is an acceptable trade-off for more punch
and transparency. If your gated bass drum sounds too artificial, then double the
track, gate one of them, and mix the other one in as ambience. A gated track can
be processed more heavily with EQs without over emphasizing the entire bass
frequency range. The open, unprocessed bass drum sound can be subtly mixed
in to reproduce the natural, ungated sound of drum sets. The lower the bottom
characteristic core frequency of the bass drum is, the shorter it should be gated in
order to keep it from being tonal, and therefore interfering with the bass. On the
other hand, a bass drum with a long tail (not heavily dampened) can just be given
a slight bass emphasis, to avoid oversaturating the bass band.
Be careful with reverb dosage on bass drums! Too much reverb brings out the
skin part of the sound in the upper frequency range and makes the transients
sound fluttery and less clear. Use reverb without pre-delay to avoid fluttering
caused by rhythmic shifting. If you do want to use pre-delay, be sure to sync it to
the rhythm and tempo of the song (for example, tempo-synched to a 32nd note
value). I like to mix in a small amount of drum kit room ambience for creating a
relationship between the bass drum and the rest of the kit. A completely dry bass
drum can very easily sound out of context and separated from the rest of the mix.
However, this effect is used in certain styles (hip-hop). The amount of reverb for
a bass drum should create the effect that it is missed as soon as it is taken away,
but doesn’t sound blatant or obvious when it is present. Large room reverb is
seldom used for bass drums since it makes the whole sound very indirect and
soft. The W2 reverb unit by Wizoo contains a very nice preset for bass drums,
making them sound bigger. If you don’t have the Wizoo plug-in, you’ll also find
good presets in the SIR II reverb. The important factor for bass drum reverb is
the quality of the early reflections of the device, especially with regards to the bass
frequencies. Unfortunately, software which is necessarily CPU optimized has its
limits, and doesn’t have the same quality as dedicated hardware. The M7 from
Bricasti is one of the high-end hardware units in this area and is a real reason to
deal with the inconvenience of external hardware.
For boosting the snare’s overtones, I recommend using a high-resolution highshelf filter or a parametric filter with a broad band Q-factor.
The snare should also be in the center.
Like the bass drum, the snare drum has two characteristic core frequencies. The
bottom is a little higher than that of the bass drum and is situated in the lower
mid-range. Snare characteristic core frequencies are between 120 and 260 Hz,
depending on head size or samples being used. Snares are also often recorded
with both a top and a strainer microphone. The top mic generally picks up the
punch in the lower mid-range and the strainer mic is used for getting the strainer
sounds in the upper mid-range to the higher frequencies. In the upper frequency
spectrum, the strainer can sound broadly noisy up to 20 kHz and beyond. When
using two microphone tracks, it is important to pay attention to phase. If there
is any phase-reversal of the strainer mic during recording, then both tracks can
cancel each other out in the mid-range, possibly taking all the power out of the
snare sound. Try this out by simply switching on the phase button on the strainer
track while listening to both tracks at the same loudness levels. I find it interesting
that with live recordings, snares first start sounding really open and good when
the overhead tracks are mixed in. Snares seem to develop their true, full high-frequency sound with a little distance, making a second mic necessary to adequately
record the sound.
If you are unhappy with your snare sound, there are two areas to look into: the
bottom and the overtone spectrum of the strainer. In addition, I recommend
using a lowcut filter at 80 Hz and also some attenuation to create space in the
mid-range. Snare sounds from metal snare drums add an additional interesting
element into play. These frequencies can be boosted or attenuated with a bell filter
set to a medium Q-factor. If the snare tone has a significant relationship to the key
of your track, then it is easier to integrate such a stylistic element into the song. If
the sound is a half tone lower, for example, it will probably sound terrible. Don’t
be afraid to pitch-shift the whole track with your DAW to tune it to the song’s
fundamental key, or to a fifth above the key fundamental. The Red Hot Chili
Peppers have tuned and integrated the tonal sound of a metal snare into several
of their songs. If the sound is annoying, you can use the track doubling technique
described below to push the sound into the background.
Compressors and Gates
Compressors fill two roles for snares:
u Top end limiting or compression to decrease dynamic range. Here you need to
make sure that the natural transients are not affected by compression, since this
can lead to a tide-drum effect (imagine a sound that a cardboard box makes
when played), sounding like someone took the air out of the snare; it destroys
the natural transients by too much dynamic processing.
u Bringing out snappiness and percussiveness with longer attack times.
u Controlling the quiet parts of the signal (strainer) by fast release times (the
faster the release time, the more strainer sound).
When working with live drums, it is standard to use a gate on the snare track or
to create a copy of the snare track to be gated.
Doubling Bass Drum and/or Snare Tracks
I have already mentioned this theme in passing when talking about gates with
bass drums. The trick of doubling (or “multing”) is very often used with tracks
(bass drums and snares, for example) with microphone crosstalk – particularly when there is only one microphone track (head/bottom or top/strainer). By
processing the doubled tracks differently, you can more intensively manipulate
the signal, without making things sound artificial. If you put a short gate on the
copied track, you can use extreme EQ and compressor settings. Strong EQ settings on short, impulse-type sounds are not as noticeable as they are on ungated
sounds. Signals freed of crosstalk can also be more heavily compressed. Strong
compression with quick release time of an ungated track with crosstalk has the
undesirable effect of boosting the quieter (crosstalk-laden) signal parts.
A track processed in this way is then mixed with the ungated, “natural” sounding
track, resulting in more punch.
This technique can also be used to split the instrument in its basic parts (bottom/
skin) and then put it back together. In this case, one track would have the low
frequencies and the other track the high frequencies.
The same method can be used when dealing with big reverbs on snares: The copied and gated snare track drives the reverb with the short, “cut” snare sounds
which are powerful and clean, and not washed out with the crosstalk signals. The
open, ungated track is then mixed in with a small amount of ambience reverb.
Certain styles demand that the snare is a bit more “meaty” by putting the gated
track in turn through gated reverb. You can create such a sound by putting a noise
gate after the reverb device. When both tracks are mixed together, the snare still
sounds natural enough. The gate release time must be synched to the song tempo
(usually to a quarter note).
Other Aspects of Drums
I like to mix drums from the listener’s perspective, so that the hi-hat is – for a
right-handed drummer – slightly to the right in the panorama. You must be
sure that the cymbals are organized so that the hi-hat is also on the right channel. This is where mix-ups can occur; one engineer might think in terms of the
drummer’s perspective and another engineer might have the listener’s perspective in mind. Track labeling can lead to misunderstandings. Make the panning
of the cymbals as wide as possible. Very wide left-right panning is usually impossible for reasons of phase problems. This is the time to watch the phase behavior with your correlation meter; make an acoustic check by switching the
control room mix to mono as well. I also spread the tom-toms from right (rack
tom) to left (floor tom) in the panorama. To bring out the authenticity factor,
the hi-rack tom should be only slightly right, the mid-rack tom slightly left, and
the floor tom farther left. When more toms are used, you can simply follow the
natural placement on stage.
Gating toms is mostly done manually (scissors tool in your DAW). With a small
amount of fills, this is accomplished rather quickly. If there is a lot of tom playing
in the drum part, then a noise gate can save you time. If the toms sound a little
weak, an extreme (good-sounding) bell filter with a medium-to-strong Q-factor
can provide the necessary punch. In this case, try to find the exact characteristic
core frequency of each tom. Depending on the style, you can also mix in a bit of
the big-room snare reverb to make the toms sound bigger.
The hi-hat needs to have a lowcut filter to eliminate any low-frequency air sounds.
In order to reduce crosstalk from the snare, you can also set the lowcut to 250 Hz.
This really depends on what kind of hi-hat is used and the style you are dealing
with. A heavy rock hi-hat definitely would need a lower cutoff frequency setting.
I have never used a noise gate on a hi-hat track; it seems rather pointless to me
that when in certain parts of a song there is no hi-hat, the artificial switching on
and off of the track would create sonic inconsistency. The hi-hat mic also creates a
kind of super-close ambience track, which contributes to the global sound of the
drums. The more deep frequencies you leave in and the more you pan the hi-hat
to the right, be aware of the center balance of the whole drumset because of the
snare crosstalk on the hi-hat track.
The cymbal track is already a kind of close-ambient room sound. Here, again,
you’ll need a lowcut filter. A gentle roll-off of 12 or 24 dB per octave is best in
order to avoid unnecessary phasing effects. Use your ears to determine the cutoff
frequency. In some situations, it serves the sound aesthetic best when the whole
frequency spectrum comes through. In other cases, a cutoff at 400 Hz is appropriate. Keep in mind that overhead cymbal mics have a 6 ms acoustic delay (because
of being a distance of 2 meters (6’ 7”)) with regards to the snare and this can result
in a slightly “fluttery” bass drum sound if the cutoff frequency of your lowcut is
set too low. If the drummer forgot cymbal cleaning fluid for the recording session,
you might want to try a gentle-sounding highshelf filter that brings some brilliance
starting at about 12 kHz. In this case, pay attention to the quality of the plug-ins
your are using. Try the UAD-1 Precision, the Helios 69 or the Neve 1073 EQ.
Simulating Ambience Tracks with
Offline-processed Convolution Reverb
Complex high-end productions get their depth from recording ambience and
early reflections in large, expensive, and flexible recording studio rooms onto
separate tracks. That makes it easier for the mixing engineer to mix in a subtle
amount of room ambience and depth, without oversaturating the recording with
room ambience from artificial reverb devices. Often, three different stereo ambience tracks are recorded – close ambience with small diaphragm mics about a
meter (3´3´´) away; a mid-ambience with big diaphragm mics about three to
four meters (10´ to 13´) away; and then boundary layer mics (PZM – Pressure
Zone Microphone) up to 8 meters (26´) away from the drums. If you don’t have
access to such big, expensive rooms, you can reach into our bag of tricks: As
usual, our drums are coming out of a sampler, a software instrument, or a re-
ally small and dry recording room. Compared to your favorite CD, your drums
might have a good amount of punch, but they still sound flat. This phenomenon
has nothing to do with the quality of the samples or the recordings, but simply a
result of a lack of room information in the sound.
Now create a preliminary stereo mix of the drums where you mix the bass drum
in much more quietly than you would in the final mix. Depending on what DAW
you are using and your general workflow, there are various methods that all have
the same objective – to create one or more artificially created ambience tracks having a large amount of room ambience, which can be mixed into the drum tracks.
One possibility is exporting and re-importing the above-mentioned drum mixes
along with offline-rendered reverb. For this you can use the internal convolution
reverb in Nuendo or any good convolution reverb in offline-processing mode. If
you do not have this possibility, you can switch the drum group to solo and insert
the reverb into the group track. Then you can export the reverbed drum mix
and use it as an ambience track to mix in. By choosing the right impulse file and
amount, you can obtain a stereo ambience track which is comparable to a stereo
ambience track made from two boundary layer mics placed far enough away
from a drum kit in a good-sounding room. The bass drum can remain latently
present in this ambient track, so that it is part of the entire sound without creating a flutter-effect. With high bass drum levels, the skin sound can tend toward
flutter, and sounds old-fashioned and “wet.” The result of this method is similar to using a good reverb device in real time. It saves you computer power and
simulates a realistic situation, depending on the reverb used. The loudness of the
ambience track or of the ambience reverb should be adjusted so that it is neither
obvious nor blatant, but is immediately missed if muted. Successfully layering
drums in discrete room ambience makes it clear what is meant by flat-sounding
Refining Loops with Doubling, Gating,
Compression, and Reverb
Not every loop that sounds good on a sample CD is also good for using in a mix.
Often loops sound “too good” and dominate the whole frequency spectrum in
a way that the mix would be perfect with the loop, if only there weren’t other
instruments involved!
How do you give the clap on a loop some reverb, without putting reverb on the
hi-hat and cabassa? How do you make the snare sound punchier, without mak-
ing the bass drum sound mushy?
This is all possible by simply copying the loop a number of times! Use each copy
of the loop for shaping one particular aspect of the sound. In this way you can
use a copy to process the bass drum, another to separate out the snare, and yet
another one for the rattling sounds. Shape the desired features with gates and
compression, with lowcut, shelving, and bell filters, and with appropriate use of
reverb and delay.
(Sound Example #08 and Exercise “Loop Editing”)
Sound Example
I have already mentioned the most important thing for a good bass sound: The
bass needs room! Every unnecessary low-frequency event robs the bass of the
room it needs for clarity. Once again, use as many lowcut filters as you have individual tracks! If you are recording your own bass guitar tracks, use fresh, thick
strings, with a high string action (unscrew the truss rod ½ to 1 turn), so that
the strings can vibrate freely with less noise. New strings ensure that the strings
start vibrating more quickly, helping intonation and presence. A good front-end
should also be used – at least a very good DI box or tube DI box (for example
BSS). If you have a UAD-1, the 1176 and/or the LA-2A in the bass channel is a
must. Here you can adjust the crispiness and therefore the localizability of the
note beginnings by playing with the attack times.
In addition, you can control the balance between the “heavy” sounding notes and
the stops, damping sounds, and dead notes with the compression amount. If the
bass is played well, the more quiet dead notes can be brought out with the compressor. If it is badly played, or if dead notes disappear completely from the mix
from time to time, you can manually cut out and mute the annoying elements.
In addition to the timing of the note beginnings, note duration is also particularly
important regarding interaction with the bass drum. This is achieved through the
player’s ability to skillfully dampen the notes. On the strong beats, particularly
with the snare (usually on 2 and 4 of a 4/4 measure), the sounds in a bass line
should be ended or stopped, so that the line grooves. If necessary, use the scissors tool in your DAW along with a short fade-out and muting to create room.
Remember our supreme aim of only including events in a mix that have a welldefined place!
The compressor is a versatile dynamic shaping tool for the bass guitar as well, and
can be used in the following ways:
u Dynamic limiting of sudden level peaks and irregular playing.
A bass EQ curve might look like this. Above the lowcut filter the bottom is boosted and the thumpy mid-range is
attenuated. Around 1 kHz, “comprehensibility” is accentuated.
Because of the low frequency content of the bass, it always belongs in the center.
With stereo tracks from synthesizers or software instruments, watch for a good
balance because an unevenly panned bass leads to a left-right imbalance resulting
from the high energy contained in the bass sound. Use a goniometer or the peak
meter in the master bus while soloing the stereo bass track.
Sound Example
A bass can only handle a lowcut filter with a steep slope – 36 dB per octave is
often not enough, with the result of dampening the useful frequencies starting at
about 30 Hz. Some basses that don’t have enough bottom end can benefit from
a boost between 40 and 80 Hz with a medium Q-factor. You can dampen the
lower, thumpy mid-range between 120 and 350 Hz if the bass tends to take space
away from other instruments. In the area between 800 Hz and 1.2 kHz you can
adjust the nasal, woody part of the sound. This also helps with tonal localization.
Above that, there are just a few very quiet overtones and mostly dampening and
other noises. In Sound Example #09 you can hear a soloed bass with and without EQ, then in the mix with and without EQ. The Oxford EQ was used for this
u Creation of sustain with long notes (particularly in songs with slow tempi).
With a long release time – synched to song tempo – the level reduced by the
compressor amplifier slowly comes back up until the next note creates another
reduction of level. This way, nice, long tones can be drawn from a lower-quality instrument.
u Boosting quieter side noises (funky or Jaco style) with fast release times (this
excludes creating sustain as described above).
u Supporting rhythmic localization and percussiveness with long attack times.
With normal pop music, you should give the bass – at the most – a subtle and
barely audible, but just perceivable, room ambience from the drum kit. This emphasizes the unity between drums and bass.
Especially fretless basses can do well with a subtle bit of room reverb. Other bass
reverb techniques fall under the category of stylistic particularities.
Multiband Compression for Bass
In particular cases, a multiband compressor with expander function can be used
to bring out specific spectral regions of an instrument. If a standard compressor
is unable to boost the dead notes enough that they can be heard in context, then I
like to use the multiband compressor from Steinberg/Spectral Design by copying
the bass track and running the copy through the multiband compressor which is
set for bringing out the frequency bands in question. In the Steinberg compressor, you can keep the solo switch on for certain bands, so that only the critical
frequencies come through, which are then mixed to the “full range” channel.
This emergency tool can also be used with other instruments to undertake even
more extreme “first aid” operations.
Doubling the Bass Track
Since the bass also consists of two components (bottom and top), it can be helpful in some cases to split it into two tracks in order to work on the bottom with
one track (using, for example, the Pultec-EQ on the UAD-1) and shape the tonal
comprehensibility of the other track. Interesting results can be obtained by doubling tracks in this way, sometimes making it possible to save a mediocre bass
Individual lead vocal tracks are usually placed in the center in pop music productions. In parts of songs that are meant to deliver a feeling of intimacy, single lead
vocal tracks are always better than doublings (different takes). If a single track
is not enough to create the presence needed, you can also duplicate the track
and decorrelate the two – in other words make them dissimilar – with various
compressors and EQs, then slightly split them between left and right. Be careful
to maintain the phantom middle placement – the impression that the lead vocal
track comes from the center. This strategy gives the voice more punch and presence. To create intimacy you can extremely compress one track (for example with
a Sonnox Inflator), make more presence with boosting high frequencies and then
mix with the main track. For pop stars who have thin voices, the critical mass is
often exceeded with vocals doubled 40 to 60 times, and then putting AutoTune
on each track along with a good reverb plug-in. After quite a bit of editing you
can also get such results and generate the impression of one full voice. Maybe you
know the phenomenon where a large number of single voices sound at one point
like one voice. That can even happen with amateur choirs, when enough voices
come together.
Bass and Chorus
Doubling the bass track is a great way to add chorus, flanger, or other effects.
Leave the bottom track without effects and put a roll-off with a soft 12 or 24 dB/
Octave highcut filter at around 180 Hz, for example. Then do the opposite with
the copied track by dropping on a lowcut with a steep slope and then insert the
modulation effect. This way you maintain a clean mono bass without phasing
problems along with a full-sounding, modulating stereo effect.
The better vocals are recorded, the easier it is for you to place them in the mix.
This can be compared to digital photos: If you take a picture of something with
a 2 megapixel camera for less than $100, you might have a nice little photo. But
if you want to make a poster out of the picture, then you quickly see the limits of
the image quality – you can easily see the individual pixels and the lack of depth.
Pixels to the eyes are bits to the ears. A recording, which does not have a high
resolution, cannot be naturally placed in a three-dimensional room. With a high
resolution, lead vocals come right out of the loudspeakers.
Even if you do not own a first-class front end chain (microphone, cable, preamp, cable, converters), then I strongly recommend that you borrow or rent such
equipment for the recording session so that you can compare the results to those
of lower-quality equipment. This way you’ll get a feeling for the reserves that are
hidden in good technology and the limits and possibilities which good engineering can provide. Good performance together with excellent recording equipment
makes good engineering easier.
In pop music, refrains are very often doubled. Four doublings per singer and
voice is generally the average. With a lead singer and three backup voices you can
easily end up with 16 voice tracks. To organize sonic separation, I recommend
routing the backup vocals to a separate group. The doubled lead vocals should be
panned in a way to keep the phantom middle position, with the backup vocals
imbedding the lead in the panorama. The backups should be panned so that the
higher voices are more spread and the lower ones are more towards the middle. Mix the group first together, before putting them into the whole mix. In the
group track, you can insert some stereo expansion in order to make sure that the
backgrounds and lead have enough separation in specific passages by increasing
the width of the backups.
All vocal tracks need a lowcut filter – 80 Hz is enough for the lead vocals and
I like to give background vocals a softer 12 or 24 dB per octave filter, setting it
higher (up to 400 Hz) depending on the style and situation.
Vocal recordings often suffer from resonance in the bass and lower mid-range
that arise from suboptimal room acoustics, which result in room modes and
comb filter effects. Since recording situations often only eliminate disturbances
in the foreground affecting the mid and high frequency ranges, the recording
can sound thumpy. The sound could also be described as being “closed” instead
of open. In these cases, the resonance frequencies need to be scanned between
120 and 350 Hz with a very extreme and steep-sloped boost in order to find
them then reduced with a notch or bell filter (with high Q). The result is singing
that sounds more free. This mid-range band must not sound too thin! Here is
where warmth is located, especially from male voices.
In the mid-range region up to 2 kHz for men and 3 kHz for women you should
boost with a wide (low Q) EQ for speech comprehensibility. A soft boost in
this range is standard, since putting the vocals in the phantom middle creates a
comb filter effect through the dampening of the human head, which weakens
these frequencies. If the signal coming out of a center loudspeaker sounds like
it’s coming from the front, then the signal comes into the left and right ear in
an equal and unified manner. If, during simultaneous playback from the left
and right channel, there are time differences between the two channels; then
the signal coming from the one speaker arrives in the opposite ear with a slight
delay and is dampened, and together with the other speaker creates a comb
filter effect.
Frequencies between 6 and 8 kHz are sensitive. Here are the “s” and other sibilant
sounds that can be found up to 12 kHz. Boosting should always be subtle and
broad-banded, while paying careful attention to the sibilant sounds. Sometimes
combining with a de-esser can help. You can create a certain openness by gently
increasing the highs from 10 to 12 kHz and beyond. For this, a shelving filter is
often used. For processing high frequencies, I highly recommend using EQs that
are gentle and internally operate on the basis of oversampling. The Pultec and the
Precision-EQ both have oversampling engines and sound very silky (UAD-1).
Sound Example
Sound Example #11 is a successful case of vocal sound first aid. You can test your
EQs with this example.
Here you can see extremely pronounced sibilants resulting from a gap between the upper two front teeth. They are
manually reduced in the vocal track. The words are: “She was so strong...”
A de-esser can be used to handle a moderate amount of unwanted “s” sibilant
sounds. There is a wide range of high-quality native and DSP de-essers, from
Waves, Steinberg/SPL, UAD-1, and TC. A de-esser is a kind of higher frequency
expander or compressor, which reduces sibilants with a bandpass filter on the upper mid-range level peaks without making the rest of the frequency range sound
dull through dynamic reduction. That is the objective of de-essing and it is possible without a pumping sound, as long as the sibilants are not too much in the
foreground and therefore must be reduced too strongly.
Another kind of de-esser is a dynamic EQ. A native version is the Sonalksis DQ1,
for PowerCore there is the excellent TC Dynamic EQ. With dynamic EQs, a causal
relationship with dynamic events is established. In other words, the frequency band
in question is processed with a bell filter when a certain threshold value is exceeded.
Threshold recognition can be set up independently of the band that is to be processed, via side chain (detector input). This opens up a huge field of possibilities.
Manual de-essing is the most effective and least damaging for the sound. In this
case, enlarge the waveform representation of the soloed track to be processed and
cut every sibilant in order to manually reduce levels. This might sound like neverending editing drudgery, but it actually can be done quickly if you consistently work
with keyboard shortcuts. Sibilants are optically recognizable by having an extreme
degree of sound density, while being relatively weak in level.
Sound Example
Making Different-sounding Takes Similar
You should reserve the best plug-ins for lead vocals. Here you can also use the
LA-2A or the 1176. The best copies are the “originals” on the UAD-1 platform,
but there are numerous imitations of these legendary devices as native plug-ins
as well. The 1176 is THE vocal compressor! Only with this device is it possible to
get the Bob Clearmoutain vocal sound (http://www.mixthis.com). When talking
about this, often the term “bite” is used, which can be controlled by using a long
attack time. This way, the beginning impulse of the vocals comes through unprocessed before the compressor kicks in. This is especially audible with plosive
consonants (Sound Example #10).
Sometimes vocal takes are recorded on different days. Along with the mental
mood and the physiological condition of the voice, the slightest variation in mic
placement or the singer’s position with respect to the mic can have a major influence on the vocal sound, so that attempts to improve things later sound quite
different. It is easiest to make takes sound similar on the spectral level. If a parametric EQ and good hearing don’t lead to the desired results, then so-called intelligent EQs may help. Intelligent EQs examine both the file to be processed and
the target file in order to make a suggestion for processing. The TC Assimilator is
such a device on the PowerCore platform. It allows you to morph between 100%
original and 100% processed. If the files are carefully read in, you can get good
results. However, fine sound variations stemming from day-dependent timbre
changes as well as differences in room acoustics resulting from different recording positions cannot be corrected with such a device.
For lead vocals you can also use the 1176 compressor with high ratios at 12:1.
Even when the compression sounds extreme in solo mode, it sits into the mix
well and “sticks to the front.”
Reverb and Delay
Most of the time, the lead vocals are placed way in front of the mix. Using a large
room reverb is a common mixing mistake because then the lead vocals sound
less direct, and it seems as if the singer has taken a few steps back!
One of the many possibilities is to use large amounts of small room reverb. These
rooms should be subtle and unnoticeable, so that the listener is not consciously
aware of them. Often these small rooms are combined with bigger rooms and delay. A well-employed delay can help make the vocals sound fuller without pushing them into the background.
In Chapter 6 (Dimension 3 – Layering with Reverb and Delay) I have already talked about this in more detail.
If spectral alignment is not enough, then you can experiment with plug-ins
that focus on vocal manipulation, such as the TC Voice Modeler or Celemony’s
Melodyne. There is also a number of voice processing plug-ins on the native
Processing Breathers
The amount of “breathers” in a track is a question of style that I cannot answer
here. But breathers that are simply cut off are not a question of style, but a question of a lack of skill! Decide for every new production how you want to deal with
the issue: Should breathers make up a major part of the vocal sound or should
they be unobtrusive? Carefully controlling this in “big productions” is essential.
When doubling tracks, I recommend avoiding breather doubling. Edit out the
breathers from the copied tracks (make sure that the breathers are placed in the
center). If some breathers are unattractive or inappropriate, then replace them
with other breathers. With background vocals I like to split two or four synchronous breathers left and right in the panorama and remove all other breathers
from the vocal tracks.
Editing Sloppy Syllable Endings in Vocals
I recommend that you devote as much attention to syllable endings like “t” and
all other rhythmically significant consonants just as you would breathers. If “t”
syllable endings chatter in the background vocals or if there is no consistency
with the lead vocals, then it sometimes helps to simply cut them off and mute. To
avoid artifacts, put short fade-outs on the edits. This way there will be a rounder
sound and the background vocals frame the singing better. Often I completely
fade-out the end consonants of individual vocals, so that the soloed vocals sound
somewhat weird. In the context of the whole mix, however, such edits are not
perceptible – to the contrary, they create greater transparence.
Because all other instruments are handled very individually according to genre,
taste, sound aesthetic, and arrangement, it is difficult to make generalizations.
Therefore I will mention the essential aspects that could be useful for your work.
Particularities of Working with Backing Vocals
If you do not have enough tracks available to simulate a large, full choir, you can
copy a few tracks and process them with AutoTune or a Voice Modeler, so that
the only thing that changes is the color of the copied tracks in order to bring more
In R&B productions 60 to 90 individually sung vocal tracks are not uncommon.
Recently, the trend has been to have up to 300 vocal tracks.
Proximity Effect
The proximity effect refers to the physical characteristic of microphones to sound
very bass-laden when they are very close to the source. The distance between
mouth and microphone is extremely small. Voices recorded with proximity effect can give the impression of “crawling out of the speakers.” It is very useful to
record a very dry signal in order to have the possibility to place the sound in the
front. Because of the close proximity of the sound source to the microphone diaphragm, there is a large level difference between the first signal and the room response. The room sounds are therefore much less significant in recordings where
the distance is 10 to 30 cm (4’ to 12”) from a large diaphragm mic. Because of
the very small amount of room sound, vocal tracks that are very close-mic’d are
easily placed in the mix and therefore can be compensated for bad recording
room acoustics. Using this recording method for lead vocals can give a feeling of
intimacy and bring out the vocals in front of the mix.
Acoustic guitars have a very wide spectral range and often take up too much
space in the mix if they are not processed in some way. Use a lowcut filter to
eliminate unwanted low frequency noises. If an acoustic guitar is placed predominantly in the foreground, a lowcut at 80 Hz is sufficient, so that there is a
good deal of warmth remaining in the lower mid-range. If, on the other hand,
you wish to have only the soft and velvety strumming (with a soft plectrum)
sound as a rhythmic accent, then a lowcut at 400 Hz can be just right, as long
as the lower mid-range is already occupied by other events. This way, you avoid
overlapping and lack of transparency of this sensitive frequency band. In this
second case, I recommend panning far from the center. In cases where there is
an acoustic guitar alone in the intro or with singing that later comes in with the
whole band, I automate the frequency parameter of the lowcut EQ so that the
cut off frequency shifts upward as soon as the other instruments start playing in
order to make room for them.
The rich overtone spectrum of acoustic guitars can often be brought out with
a high-quality EQ. In this case, a shelving filter starting at 8 kHz is appropriate. Remember that a large amount of high frequency content psychoacoustically
suggests closeness!
Use the Compressor Exercise #07 to get familiar with the many uses of compressors. Since acoustic guitars are very dynamic, a top quality compressor can help
narrow down the dynamic range. With fast release times you can bring out the
various playing noises, so long as they have musical significance. Lengthening attack time makes the guitar more percussive. Long release times even out the level
of loud sounds that naturally have little sustain and lose their loudness quickly
(Exercises 01 and 02). Uncompressed guitars are difficult to fit into complex mixes. The typical level differences created while playing are often too big. The sonic
luxury of natural, uncompressed guitars is only for puristic mixes with small instrumentation and skillful performance.
Reverb, Delay, and Chorus
Use reverb and delay for all types of layering. Chorus is often used for acoustic guitars. The detuning effect of chorus gives 6-string acoustic guitars a more
“floating” sound, similar to 12-string guitars.
If you have any influence on the recording process, make sure that fresh strings
are used. Also, verify the combination of microphone and pickup recording
methods. Using only pickups does not sound as natural as micing. The ceramic piezoelectric pickups often built into acoustic guitars sound a bit artificial
and hard. Spend time playing around with the microphone position until you
have found the optimal placement. Sometimes just 5 cm (2´´) here or there can
make a huge difference! If the acoustic guitar is playing a major role in the arrangement, it is worth looking into M/S micing. You’ll find more information
in technical literature on mics and microphone techniques.
It is very difficult to make blanket statements because of the variety of sonic possibilities of the electric guitar, even though I am myself a guitar player. Modern
modeling techniques (POD and others) have made it relatively easy to get good
sounds down “on tape.” This way of working usually leads to recording stereo guitar
tracks, which in turn lead to inconsistent panning in the mix. Do not be tempted
to place guitar tracks in the center just because they are stereo, so that the full stereo effect can be heard. A good, solid mono track is usually a better basis for a good
sound; a stereo guitar that doesn’t hold up under mono because the stereo effects
are no longer there to support it is questionable. I often find it more interesting to
work with the combination of a single dry guitar track recorded simply through
a DI box along with a mono guitar track with the amp sound – whether recorded
with physical amps and mics or modeled. This way, you have an amp track without
effects giving you sound information as well as a dry track, which gives you the
possibility of refining this sound conception with other amp simulations, or real
amps that can either augment or replace (re-amping) the amp track. By recording
in this way, you can at least rectify possible recording mistakes. I believe that the
many possibilities of the numerous amp simulations are only really useful when
you have intensively worked with – ideally – the original amp models. Often using
presets leads to mediocre results because the preset and guitar don’t always match
100% with each other. When looking for the right sound, keep the effects off so
that when you do find the right sound, the effects are the icing on the cake.
If you value good electric guitar sounds, I recommend seriously studying amp
micing techniques. Sometimes 2 to 3 cm (1´´ to 2´´) makes the all the difference
in the world with guitar sounds. The effort is certainly worth it, providing results
that still leave amp modeling in the dust.
Guitar sound freaks might want to look into the SPL Transducer. This analog
speaker and microphone simulation gives a powerful, latency-free guitar sound.
The only rule I can give as far as EQs are concerned is that using lowcut filters is a
must! Guitar sounds should be recorded as close as possible to the “target sound.”
EQs allow small corrections, but won’t work wonders. With increasing amounts
of distortion, compressors become less necessary. Distortion leads to a natural
kind of compression. On clean guitar sounds, use compressors as you would on
acoustic guitars. If your sound isn’t right, experiment with all amp modeling possibilities, including for example the Leslie from the B4II of Native Instruments.
Such experimenting can easily lead to interesting results.
Dealing with keyboards in a mix is similar to electric guitars because of the infinite variety of available sounds: watch out with stereo tracks so that you place
keyboards as far from the center as possible. Strictly follow your mix strategy
with panning. Always use lowcut filters – even the best samples can have DC
Offset and sub-bass artifacts.
HE SHORT HISTORY OF PROFESSIONAL computer-based production platforms has been characterized by quick, uninterrupted development and continual improvement of all plug-in types. This has led to
an almost unfathomable saturation of the plug-in market. In the following descriptions, I therefore can neither provide a complete picture of what exists nor
guarantee that there isn’t now a new, better plug-in somewhere in the world. My
focus is on plug-ins that work on the DSP devices that I have already mentioned.
Additionally, I will mention a few native-based plug-ins that do not use non-native DSP units. With native plug-ins, I do not mean it negatively when I say that
they require a lot of processing power. When this is the case, then the price we
pay is simply more offline processing steps. Some of the plug-ins mentioned are
included on the DVD as demo versions. If needed, check on the Web sites of the
corresponding manufacturers to see if there are more recent versions available
for download.
With the variety of available plug-ins, I do not want to create a buying frenzy nor
give you a feeling that you are lacking a specific type of plug-in. My aim is rather
to incite you to critically examine individual components in terms of quality and
to put together a small number of tools that you have not only mastered how to
use, but to do so at the right time and place. In the end, you do not need any more
than a few dozen plug-ins that you can use efficiently by knowing them well. Take
the time to separate the wheat from the chaff in your plug-in folder so that in the
future you only use devices that lead to satisfying results!
Often in online forums I am criticized for not talking about native or freeware
plug-ins and that my methods require an investment in DSP hardware. My first
concern is to show you how to achieve professional results using your DAW. I
cannot justify ignoring high-quality plug-ins simply because they usually require
more computing power, and cannot easily be developed for free. Compared to
the tremendous cost of analog equipment – which in the past was a requirement
for good sound – today’s investment is really quite small. In the past, you could
hear which tape format was used in the production; today you can hear if external DSP power is used or not.
Sound Example
You can use Sound Example #11 to test your EQs. Insert your various EQs one
after the other and find a good setting for one of them. Then transfer this setting
onto the others and compare the sound of your plug-ins using the bypass switch.
In my testing, I have used the snapshot function of the Pinguin graphic analyzer
and pink noise in order to create the closest possible EQ curves between the plugins, also for those with limited parameter choices. I put the pink noise through
the first EQ that is already set, then froze the frequency curve on the metering,
and then adjusted the other EQs to this curve.
The most aurally apparent difference in quality of EQs can be heard with extreme
high-frequency boosting. The palette goes from harsh to silky-soft. All other
differences in sound that go from subtle to distinct can be described with very
flowery language. Decide for yourself, what is pleasant and natural and what is
artificial, harsh, and sounds like filter noise. With strong bell filtering, you can
hear the various sound characteristics of the different “circuitry” in the different
EQ plug-ins. In the following descriptions, I discuss the most visually and aurally
apparent characteristics of each plug-in.
Universal Audio Cambridge-EQ
Universal Audio Cambridge EQ (UAD-1: optional)
The 5-band full parametric Cambridge EQ is very similar to the Oxford EQ
from Sonnox for PowerCore in terms of sound character. The difference lies in a
slightly different focus: it is flexible with regards to the cut and shelving filtering.
There are 19 different filter variations in the cut filter section to choose from, for
example Bessel and Butterworth. The E6 lowcut stands out from all other lowcut
filters that I know, making it a very valuable tool for removing unwanted low frequency noises, especially when the desired signal is itself very low. The extreme
slope of the E6 is paid for with phase shift and therefore can tend to color the
sound, so that you should consider carefully which lowcut filter is best in a given
situation. All five bands can be switched into shelving mode, giving you three
different modes for different resonance peaks (overshoots). This analog-inspired
EQ is a flexible high-class multi-purpose tool.
The Precision Equalizer creates silky highs through internal oversampling.
Universal Audio Precision EQ (UAD-1: optional)
The Precision Equalizer works on the basis of four times oversampling and is
actually meant to be a high-end mastering EQ. However, you can still use it for
specific mixing situations. To achieve the oversampling, the sampling rate of the
input signal is multiplied by four, in order to carry out high-resolution modeling. It is then reduced again to the output frequency. This is the recipe for soft
and brilliant high frequency boosting and is –at the same time –the source of the
Precision EQ’s prestige.
The small disadvantage of this EQ lies in the incremented frequencies. This is
tolerable, since it is not made for precise removal of unwanted frequencies. The
Precision EQ is excellent for brilliance polishing (high frequency boosting) for
group tracks such as background and lead vocals, and drums in order to gently
bring out the highs and to shape the whole sound.
The Pultec is a magical sound improvement tool.
Universal Audio Pultec & Pultec Pro
(UAD-1: optional)
The Pultec also operates internally with oversampling and therefore sounds velvety. It is conceived for mastering use. However, its excellent sound characteristics make it a valuable tool for individual instruments as well. The bass bands are
good for giving bass drums the kick they need, while the mid-range helps improving speech comprehensibility in vocal tracks, as well as giving snare strainers
more brilliance. The controls are different from standard EQs – that is why I don’t
recommend using the Pultec completely intuitively. The user manual is very well
written and empowers the reader to use this exceptional tool to its full potential.
The Pultec has a frequency range for boosting basses and attenuating the thumpy
mid-range, a range for mid-range boost (vocals), and a seldom-used attenuating
possibility for attenuating the 5, 10, and 20 kHz frequencies.
The particularity of the Pultec is its soft character. Even strong boosting doesn’t
produce a “filter” sound. Thanks to the 32-bit floating point resolution and a lack
of output level control, this EQ tempts you to boost very generously. That is perfectly ok, as long as you do not use a PowerCore plug-in downstream. In this case,
you must have some kind of level reduction before going into the PowerCore
plug-in. Alternatively, you can simply work with the necessary headroom by reducing the levels of the material you are working on accordingly, or reduce the
input level with the channel trim knob.
Along with using the Pultec for mastering, it is also good for vitalizing vocals and
pumping up weak bass drums. The bass boost at 60 or 100 Hz is very special and
gives powerless drums the foundation they are lacking.
The UAD-1 Pultec Pro is a Pultec that is augmented with an MEQ-5 – a three
band mid-range EQ. Two bands are for boosting fixed frequencies and one band
is for attenuation. The MEQ-5 is useful for boosting the upper mid-range at 4
kHz and at the same time the air band. Here, the basic version is missing a band.
Neve 1073 Channel Strip EQ
Universal Audio Neve 1073 EQ (UAD-1: optional)
The 1073 has a lot of sound for so few controls: the input level control on the
left is followed by a highshelf knob with a range of plus/minus 18 dB at 12 kHz.
The semi-parametric adjustment of the mid-range has a frequency-dependent
Q-factor that increases with frequency. The low shelving filter is followed by a
lowcut with a slope of 18 dB per octave. Five mono EQs need the power of an
entire UAD-1 card, while the more economical SE version allows 16 mono EQs
per card. This EQ is good for velvety-soft and, at the same time, discrete highs:
vocals, acoustic guitars, strings, and a lot more. It really peps up dusty or mediocre recordings.
Channel strip EQ: the UAD-1 CS-1
Universal Audio CS-1
(channel strip on the UAD-1 card)
The basic version of the UAD-1 Channel Strip EQ is lavishly outfitted with 5 parametric bands. The bands can be switched to cut and shelving filters. The results,
in terms of the openness of the sound, are not quite as good as the Cambridge or
Oxford EQs, but is still warm and pleasant. The CS-1 is an excellent basic EQ.
Neve 1081 Channel Strip EQ
Universal Audio Neve 1081 EQ (UAD-1: optional)
The Neve 1081 channel module was introduced a few years after the 1073 and
provides for more precise manipulation with legendary sound quality. The parameters are self-explanatory. You can open three instances of the 1081 on a
UAD-1 card, compared with eighteen instances of the 1081-SE. Offline processing is worthwhile here. The Neve collection from Universal Audio is certainly
one more step from large consoles to “Internal Mixing.” The 1081 gives your mix
an authentic Neve character.
use the “E-button” in the EQ section. The filter section is generally very flexible
and sounds pleasant, soft, and British. For this test I mixed an album with the
Duende. That certain British SSL charisma is pleasant and full of character in the
final mix.
SSL Duende EQ and Dynamics Channel Strip
SSL Duende EQ and Dynamics Channel Strip
(included with Duende)
As already mentioned in Chapter 2, the Duende makes it possible to use 32 mono
or 16 stereo channel strips. Any mono/stereo combination is possible, since each
individual channel is assigned to one of the 32 DSP slots. The EQ section features
high and lowpass filters with 18 or 12 dB/octave slopes and four parametric filters.
The lower and upper bands are shelving filters and both mid-range bands are fully
parametric filters. The shelving filters can also be switched to a bell filter, but have
no Q-factor adjustment. You can determine separately whether the EQ and Filter
come pre or post dynamics, or route them on the side chain path input of the dynamics. This way, you can remove unwanted signals prior to the compressor as
In the default setting, the EQs have the characteristics of the SSL G-Series models.
The Q-factor (filter width) changes in relation to gain and therefore allows gentle
manipulation, as is usually needed for correcting voice or instrument tracks in
the mix. As filter width increases with gain, you cannot carry out extreme operations or corrections. The “E” button – for the E-Series characteristics – can
be switched to decouple gain and filter width so that the unit can be used for
the familiar sterile-surgical operations that the SSL E-Series sound is known for.
For steep-slope removal of disturbances or resonances and surgical operations,
Basis-EQ PowerCore
TC EqSat (included in PowerCore)
The EqSat is part of the standard plug-in equipment of the PowerCore platform
and is conceived for working with stereo material. Separate left-right calibration
is possible, which is very seldom necessary in the mix. There are three parametric
filters and two cut filters that unfortunately have a slope of only 12 dB per octave.
Operation is a bit cumbersome and there is no shelving filter. Other than these
small limitations, the EQ sounds very good and is above all useful with group
section. Since such complex devices tend to be unintuitive, I recommend first
getting to know the dynamic EQ section only at the beginning. This highly complex EQ is a top-level tool for difficult situations and makes you really want to
experiment! The Dynamic EQ is very good for processing the sum and for mastering. Although not an everyday EQ because of its complexity, the Dynamic EQ
is certainly a special tool for particular situations.
Another special feature of the Dynamic EQ is a switchable – although CPU-hungry – linear phase circuit.
The Dynamic-EQ – a special tool
TC Dynamic EQ (PowerCore: optional)
The Dynamic EQ represents a new and very interesting type of device in the
plug-in world. A dynamic EQ is driven by a multiband compressor and can boost
and attenuate depending on the dynamics of the incoming signals. For example,
attenuation might occur only at a particular spot in the song where the frequency
band in question exceeds a certain threshold value. This is useful when a static
attenuation would make the sound too flat or a continuous boost would create
too much emphasis outside of the problem spots. A switch lets you reverse the
operation mode, so that the EQ stays on as long as a given threshold is not exceeded. Dynamic EQs are inspired from de-essers, which only reduce the sibilant
frequencies when a certain threshold level is exceeded. The Dynamic EQ applies
this technique to a four-band parametric EQ, where each side chain can be individually decoupled. This makes it theoretically possible to reduce the highs as
soon as a particular level is crossed (in either direction) on any other frequency
band. Static EQ values can also be set which are decoupled from the dynamic
The VoiceStrip from PowerCore
TC VoiceStrip (included in PowerCore)
The VoiceStrip is also a part of the standard plug-in armada of the PowerCore
and is made for vocals processing. The lowcut filter is very effective. The EQ section is made of semi-parametric basses and mids (no adjustable Q-factor) as well
Sound Example
as a high knob. This is a very spartan parameter set, but allows you to obtain the
desired warm-sounding results quickly. For Sound Example #11, which requires
a great deal of intervention for saving the recording, the VoiceStrip does not offer
enough fine processing possibilities. The dB display in the Lo-Gain setting seriously contradicts the values that are entered. In such cases, a simple numerical
readout from -10 to +10 would be a less irritating solution.
in over display never lights up. The Oxford EQ gives you a
choice of four different filter types that are chosen for all 5
Sonnox Bit-Depth
frequency bands at the same time. The default setting (type
lease note that the Sonnox versions
1) sounds very digital and analytical because of the decoufor ProTools and PowerCore are
pling of gain and EQ, inspired from the SSL4000 series EQs
working in 24-bit fixed point and
(for “E-Sound”, see SSL Duende channel strip above). The
the native versions in 32-bit floating
type 3 setting establishes a dependency between gain and
point (both in double precision). (Cross
Q-factor (well-known in the analog world), and is inspired
reference): Read more in Chapter 2 under,
from the character of Neve filters. The type 4 filter setting
Incompatibility Between UAD-1 and
is so gentle that it is not appropriate for any precise operaPowerCore.
tions, but is well suited for boosting highs, for example. The
sound is pleasantly silky and warm. For vocals processing I prefer the type 3 filter
because it provides a particularly pleasant and open sound. The cut filters can
be adjusted in 6 dB steps to 36 dB per octave and have an overshoot in the cutoff frequency area that is controlled with the Q parameter. For a standard EQ,
the lowcut is good, but does not attain the slope of the E6 lowcut filter of the
Cambridge EQ. The LF and HF bands can also be used as shelving filters with
various controllable overshoots.
The Oxford EQ on PowerCore is a flexible general-purpose EQ
Sonnox Oxford EQ
(optional PowerCore and native)
The Oxford EQ from Sonnox (formerly Sony) is adapted from the famous
Oxford mixing console. Like all PowerCore plug-ins, processing is at 24-bit resolution. This means that it is sensitive to overs coming from 32-bit signals. You can
solve this problem by using level reduction upstream, so that the internal plug-
The classic among general-purpose EQs
Waves Renaissance EQ (REQ4)
Since the Renaissance bundle is widely used, I have chosen the REQ4 with four parametric bands as an example of a channel EQ. Bundled with smaller and bigger EQs to
choose from, the CPU power requirements can be matched to the needs of the
various bands you need to deal with. Included are low and highcut, bell, and
shelving filters to choose from, making the REQ a universal tool. In the high
frequency ranges there is a noticeable difference between the Channel EQ and
the REQ, but still lacks the velvety character of an Oxford filter. Here there are
prehensible frequency band limitation of the bands to choose from. If the lowest
band is already used for a lowcut filter, the second lowest band only goes down
to 258 Hz. This is often not low enough to eliminate problems in the thumpy
lower mid-range. Also, the idea of copying legendary passive mastering EQs with
incremental frequency settings is an unnecessary limitation for this EQ, which
otherwise is very good.
Waves Linear Phase EQ
Neutral-sounding high-end EQing with the phase linear PEQ-Orange by Algorithmix
Waves Linear Phase EQ (LinEq Broadband)
Algorithmix Linear Phase PEQ Orange and Red
The advantage of linear phase EQs is their neutral sound. Analog filters always
have a slight phase shifting because of their physical structure, leading to desirable
or undesirable sound coloration, which can make up part of the sound character
of a particular EQ circuit. Only in the digital domain is it possible to go beyond
these physical limits and to build phase-neutral – and therefore sound-neutral
– EQs. Because of the large amount of CPU power required, it makes sense to use
such EQs in groups or in the sum and only by extreme EQ-needs as an insert in
an individual track. When standard EQs create unwanted phase effects through
strong boosting or attenuation, this kind of EQ is a good choice. The huge choice
of special filters also suggests many uses. One clear negative factor is the incom-
Beneath the Waves Linear Phase EQ it´s worth mentioning both of these excellent 10 band linear phase EQs from Algorithmix. They belong to a small group
of professional native high-end EQs, and to a lesser degree to analytical and neutral sounding tools, which allow working in the ultrasound regions with internal
sampling rates of 384 kHz. The range of uses of these special EQs include – other
than mastering – extreme processing of single bands from individual tracks without the side effects of the sound-influencing character of otherwise typical phase
effects. For example, more extreme bass boosting or subtle notch filtering is possible than with standard EQs.
Native all-rounder: the SC-517 Mk2 by Sonalksis
Analog charm with the PEQ2A by Nomad
Sonalksis SV-517 Mk2
Nomad PEQ2A
Nomad offers a wide range of special EQs. The PEQ2A is a three band vintage
EQ emulation of the Blue Tube series, which attained astonishingly good results
in my testing. The PEQ2A is not suitable for surgically precise operations that are
possible with parametric EQs. Instead of limited parameter choices, the user is
rewarded with a powerful and assertive sound. The EQ sounds very warm and
natural. The Nomad PEQ2A is therefore a native alternative to the Pultec EQ on
the UAD-1 platform. The PEQ2B is even more similar to the Pultec, while the
PEQ2C is similar to the MEQ5 of the Pultec Pro.
The Sonalksis SV-517 Mk2 is a real native alternative to the Sonnox Oxford EQ. It
has five fully parametric bands, with both the lowest and highest being switchable
to lowshelf or highshelf, in addition to a highcut and lowcut filter. Three variations are available for bell and shelving types. Once you have made the proper
adjustments, the result is so similar to the Oxford EQs that it is practically impossible to hear the difference. The Mk2 version also has a spectrum analyzer. All in
all, it is a very analog sounding general-purpose EQ.
EQ: Conclusion
Even small nuances can make a world of difference in the sum. Choose your EQs
accordingly. Along with the difficult to describe details in sound, there are two
main characteristics that stand out with regards to EQ sound: the quality of being
able to produce silky highs on the one hand, and on the other, the sound of a bell
filter with strong boosting. The latter is important when you need to emphasize
the bass drum or tom, without creating an unpleasant filter sound. The phase
problems of EQs in the mix are often exaggerated and are no problem in reality.
Phase changes can even create pleasant side effects and create a certain charm.
Phase linear EQs are only appropriate for neutral-sounding material that should
be manipulated as little as possible. This may be the case when dealing with pop,
jazz, or classical acoustic recordings or when it’s necessary to strongly boost while
keeping a neutral sound. Generally, in order to cover all processing requirements,
we need a workhorse EQ for basic work (for example, the channel EQ in your
DAW), a universal parametric EQ with cut and shelving filters (for example, the
Cambridge, Oxford, SSL, or Sonalksis EQ – there are of course other creators of
good EQ plug-ins), a tube EQ like the Pultec, and a phase linear EQ. With four
good plug-ins – which you know well in terms of sound characteristics and filter
types – you can master the frequency domain.
Basic tool: Nuendo Channel EQ
Nuendo Channel EQ
Sound Example
Using Sound Example #11, a direct comparison between the Oxford or Precision
EQs and the efficiency-optimized Channel EQ clearly shows the weaknesses of
the Nuendo plug-in. The highs are harder, and the vocals don’t breathe and do
not have the assertiveness they need. Nevertheless, the Channel EQ does its job
quite well, making it useful for tracks that are less important than the lead vocals.
The more extreme EQ settings and the more important the track, the more I
recommend using “external” EQ plug-ins. According to Steinberg, the Channel
EQs of Nuendo 4 should sound much better, but I have not yet had the chance
to test them.
Judging and comparing compressors is much more difficult than EQs because
of the many different needs these devices fill. The “normal” single band compressor – with its usual basic parameters threshold, attack, release, and ratio
– should be able to handle all the jobs described in the section on Compression
Exercises. Another category is the “beautification compressors” like the LA-2A
or the FA-770, which make the signal not only louder, but also particularly more
present and pleasant through discrete tube circuitry. I have purposely left out the
category of multiband compressors. They belong to mastering and not mixing.
Limiters are important to mixing, since for this purely technical task, it is important to be able to use a good-sounding and properly functioning plug-in which
tends toward coloration as late as possible. Like multiband compressors, loudness maximizers also belong to mastering. In these times of over-compressed
mixes, I would nevertheless like to mention this category. Finally, I will mention
the Sonnox Transient Modulator and the SPL Transient Designer on the UAD-1
platform, which – as far as I know – provide unique transient modulation possibilities in DAWs.
Please use the Compression Exercises with your own audio files to test your compressor plug-in repertoire in order to weed out and optimize what you use, so
that you have a small choice of reliable devices which you completely master.
Universal Audio 1176 Compressor – the Original
Universal Audio 1176 LN (UAD-1: optional)
I do not need to say much about my favorite compressor, since I’ve already heavily praised it in this book. It is a good choice for nearly all compression jobs.
The 1176 is well known for its “bite,” which is created by emphasizing transients.
With bass guitars and bass drums, it can sound a little thin in the bass and lower
mid-range. In that case, another compressor should be used. If you want to create
extreme sustain effects (bass sound example), you would be better off using the
Sonnox Dynamics compressor.
The Neve 2254 Compressor emulated on the UAD-1 platform
Universal Audio Neve 33609 Master bus
Compressor/Limiter (UAD-1: optional)
The emulation of the classic British Neve 2254 compressor of the 1970s is impressively authentic. In addition to compression, it allows you to sweeten up the
sound the non-linear distortion, also known as warmth.
The cost of such a close copy of the original is extremely high DSP power usage. You can only use one single stereo plug-in per DSP card. For this reason, a
DSP-optimized “SE” version is also included which lets you use up to 8 stereo
compressors per channel. In this case, you have to do without the details of the
charming distortion and the internal oversampling. It is worthwhile to resort to
offline processing in order to fatten up individual tracks such as lead vocals or
the drum group with some analog warmth. As long as you follow my advice and
do not over-compress the mix, so as to leave further dynamic processing to the
mastering engineer, you can use the 33609 perfectly well as a summing compressor. Single band compressors are not designed for material that is already heavily
compressed; the result quickly tends towards “pumping.”
Here are two tips for users who do not like to read manuals: the signal path of the
33609 does not follow the front panel design from left to right, but starts instead
– logically – first at the compressor on the right, whose output level manipulates
the threshold of the limiter on to the left. At the lower right is a three-position
toggle switch labeled with the numbers 22, 18, and 14. Since digital emulations
of analog devices does not allow the compressor to be operated when “hot” in
order to create the desired distortion artifacts, this switch gives you the choice
between three different level settings, which simulate the compressor’s behavior
at three different input levels. With 22, the compressor is “hot,” while 14 is better
for a cleaner sound.
TC 247C – the 1176 from PowerCore
TC 247C (1176 emulation/included with
An 1176 emulation is also part of the standard configuration of the PowerCore.
The compressor accomplished the job of bringing out quiet sounds with fast release times right away. However, the famous bite of the 1176 seems weaker to me
than the “original.” It is rather unsuitable for the important job of emphasizing
transients. Don’t be fooled by the fact that the compressor is louder in the default
setting than in bypass mode. It would also sound better when the level is simply
increased because “louder” always sounds better. The 247 creates longer sustain
similar to the 1176, but tends to “pump” more quickly.
Sonnox Oxford Dynamics – intuitive all-rounder
Sonnox Oxford Dynamics
(optional PowerCore and native)
The dynamic section of the Oxford mixing console is a real all-rounder. Along
with the “normal” mode, the compressor also features classic and linear release
curve modes; release curves can be an important influence on sound character.
The classic mode simulates the famous and widely-used dbx 160 compressor,
whereas the linear mode provides an unusually linear time constant for experimental uses. The compressor section also features a soft-knee function that can
be adjusted in 5 dB increments for discrete pump-free compression.
The additional dynamic modules consisting of gate, expander, limiter, side
chain EQ, and warmth make the Oxford compressor a universal dynamic tool.
Especially the gate exceeds all other plug-ins of its type with attack times adjustable down to the nanosecond, reaching the quality of a Drawmer DS 201 noise
gate, which used to be necessary equipment in every studio or FOH rack.
Waves C1 Compressor / Gate
Waves Renaissance Compressor RComp
Waves Renaissance Compressor – RComp (native)
Even before the PowerCore and UAD-1 platforms came out, the Waves
Renaissance Compressor was an insider tip in terms of native compression. It
does its job well and is very flexible thanks to the switchable Auto Release Control
(ARC) and the selection between electro and opto modes.
Waves C1 Compressor/Gate (native)
This variation of the extremely flexible C1 plug-in should actually be used as a
standard insert compressor. For that purpose, however, it is a little too flexible
and unintuitive unless you have really studied this tool intensively. The C1 allows
dynamic processing in all thinkable (and unthinkable) directions and combines
a compressor, expander, gate, and side chain EQ into a universally networkable
unit. A C1 can therefore act as a dynamic EQ, an excellent noise gate, a de-esser, a de-hisser, or a bass enhancer. A standard run-of-the-mill single track compression seems nearly trite here. You must read the manual if you want to know
how to use the C1 effectively, explaining clearly the extensive function set. In the
Compression Exercises, the snare is a good way to practice dealing with crosstalk
from the bass drum and hi-hat using a large palette of dynamic processing possibilities with and without gate and side chain. Use the C1 only after you have
become very familiar with effectively using its basic parameters.
Voxengo Marquis Compressor (native)
Like the Waves C1, the Marquis compressor sports a large number of sometime
unusual parameters, which makes reading the manual rather necessary. Despite the
variety of different parameters, you can accomplish many tasks quickly and intuitively. The extra functions provide extended processing possibilities for shaping the
release curve for fine-tuning the whole sound. There is also a phase linear mode.
The Marquis covers all the details of many devices in one and lives up to the term
“universal compressor.”
Nomad CP2S Blue Tube Compressor
Nomad Blue Tube Compressor CP2S (native)
The CP2S is reminiscent of the 1176 in terms of its transient behavior, even
though the 1176 is a no tube device, as the Blue Tube emulation implies with its
name. The sound is particularly warm and smooth. The CP2S handles all typical
channel compressor tasks perfectly well. It allows both peak and RMS oriented
compression and is very intuitive.
Sonalksis SV-315 Mk2
Sonalksis SV-315 Mk2 (native)
Being so close to Neve, it is no wonder that the Liverpool-based plug-in factory
Sonalksis offers a professional channel compressor. The SV-315 Mk2 accomplishes all tasks very well. It almost seems like the programmers of modern compressors all collaborate: the SV-315 Mk2 features new, cryptic parameter labels
like “Crush.” With crush and hold, the saturation and distortion behavior, and
therefore transient sound behavior can be influenced. Type I and II represent the
two sound characters “vintage” and “modern.” The SV-315 is a serious alternative to DSP solutions, because of efficient use of CPU resources. According to
Sonalksis, you can run around 50 instances on a 3 GHz P4 computer.
Voxengo Marquis Universal Compressor
The legend lives on: The UAD-1 LA-2A is extremely close to the original
Universal Audio LA-2A (UAD-1: optional)
The hardware version of the LA-2A is a jewel for magical sound sweetening that
can be found in almost every respectable studio. Owners of the rare original can
be particularly proud. Since Universal Audio started producing the LA-2A again,
thousands of the devices have made their way into studios. There is a long history behind the LA-2A. It quickly became a studio standard soon after it was
introduced in the mid 1960s. When production stopped at the end of the 60s,
there were devices with three different labels on the market, including Urei and
PSP Vintage Warmer
PSP Vintage Warmer (native)
The PSP Vintage Warmer is now an old hat and belongs to the first of the serious
native compressors and tape saturation simulators. The Vintage Warmer simulates natural tape saturation, which tends towards limiting at higher tape speeds.
This is why “ratio” is also labeled “speed.” Despite its good sound characteristics,
the Vintage Warmer is not suitable as a standard, utilitarian compressor, but as
a tool for special situations. If you want to bring the sound events forward and
value discrete compression, then the Vintage Warmer is a good choice; slight to
medium distortion can also be generated by the device. I also recommend reading the manual in order to effectively use the Vintage Warmer. It is well-suited for
group tracks such as drums.
The LA-2A is known for its warmth and assertiveness and is often used for vocals, guitars, drums, and bass. The user can choose between limiting, and a rather
gentle compression-oriented ratio curve. After adjusting the amount of compression (peak reduction) desired, the output gain is then set.
The UAD-1 LA-2A is an excellent emulation, with the charm of the original device
superbly translated into 0s and 1s. It falls into the category of “magic toys” also has
its place in the fully digital studio and often is decisive in determining whether a
mix has warmth. It is always interesting to see how the LA-2A often makes critical
improvements in some situations and in others has no effect whatsoever. That is
part of its magic.
Magneto (native, internal Cubase and Nuendo)
The Magneto is also not a compressor in the modern sense of the term. The
Magneto is good for bringing in a bit of “analog” life to the digital environment
here and there. It also does not have the same degree of assertiveness as the LA2A. However, it is an easy-to-use sound improvement tool. It is well suited for
group tracks.
Native tube warmth as an alternative to the LA-2A: Nomad FA-770
Nomad Blue-Tube Vintage Compressor FA-770
I have chosen the Nomad Vintage Compressor because it is a native version belonging to the category of “magic toys” described above. The FA-770 is also warm
and assertive, but tends to distort earlier. The “magical” presence created by the
LA-2A is not as strong in the FA-770. Nevertheless, it is a good alternative.
SSL Duende EQ and Dynamics Channel Strip
SSL Duende EQ and Dynamics Channel Strip
(included with Duende)
Magneto Tape Saturation by Steinberg
The Duende channel strip contains a simple, but effective compressor. The “peak”
switch lets you choose between peak and RMS modes. With peak mode you can
get more extreme compressor effects. Unfortunately, manual attack time adjustment is somewhat limited. The attack setting “Fast Atk” represents 3 ms of attack
time, while when “Fast Atk” is deactivated, the auto function varies the attack
time between 3 and 20 ms, depending on the audio coming in. Surprisingly, it
is still capable of bringing out transients from vocals or toms like an 1176. All in
all, the SSL channel compressor creates a thick, analog sound – typically SSL. The
gate is also very clean and easy to adjust. Overall, the Duende Channel Strip is a
plug-in which is full of character.
Only in the past few years have audio engineers become conscious of the importance of staying under specific peak levels. People started to realize that digital
recordings, which are too hot, lead to a large number of artifacts during D/A conversion as well as data reduction (DTS, AC3, MP2, MP3, etc.) processing. Since
digital fixed-pointed calculation can only deal with values up to full scale, only
virtual overs can be recorded, consisting of a chain of full scale bytes. Mastering
professionals consider three full scale bytes in a row as an over. Even this conservative value does not guard against so-called “interleaved sample overs,” which
can only be effectively prevented when the signal goes through a higher sampling
rate (oversampling) in the limiter. This is exactly what a brickwall limiter does.
A brickwall limiter is therefore placed in the master bus, before any dithering occurs and after all other processing steps. It also makes sense to keep an eye on the
operational range of the limiter. If more than just a few individual peaks need to
be reduced, then the input level should be reduced, since the limiter should only
have a technical, protective role. If the peaks (transients) are capped too much
or too often, then the result is unwanted loss of dynamics as well as undesirable
sound transformation.
SSL Duende Stereo Bus Compressor
SSL Duende Stereo Bus Compressor
The Bus Compressor is like glue for group tracks or the summing bus. If the
mix does not really seem unified, carefully using this device can bring the whole
sound together in a natural way.
It is different if you are using a limiter to reduce strong level peaks in order to
increase the amount of the overall loudness reserves. Here we can separate the
men from the boys because lower quality limiters create artifacts and overs,
which lead to a sonic distruction of the transients. For example a punchy snare
suddenly sounds like a cardboard drum when over-limited.
In order to test limiters for technical reliability and sound neutrality during
heavy limiting, simply take the snare sample from the Compression Exercises
and normalize it to 0 dB headroom. Insert the limiters you want to test in the
master bus and switch on the first one. It should begin its work at minus 0.3 dB.
Now increase the level of the snare signal until the limiter lets an over through.
You can check the overs either on an external measuring device (for example,
PAS-Products or Pinguin Metering), the metering built into your host application (overs display), or with a level meter inserted in the signal flow behind the
limiter. Good limiters hold up even under merciless overdrive attempts, whereas
others let overs through. After this purely technical test, let’s turn to the sound:
increase the level until “there’s no more there” in other words, until you have the
impression of hearing unnatural transformation of the sound of the transients.
A good limiter should have attack and release times that are so well adjusted that
transients are deformed as little as possible.
The excellent Sonnox Oxford Limiter goes a step further: if desired, it makes a
new transient interpolation. This is an ingenious function that restores cut off level
peaks. After this little experiment you know which brickwall limiter you will use in
the future.
TC MD3 Brickwall Limiter
(PowerCore: optional, in the MD3-Bundle)
The technical and sonic superiority of the MD3 Brickwall Limiter from PowerCore
is the undefeated champion of the TC Electronic company. Like the UAD-1 limiter, it cannot be overdriven, but tends to deform transients much later. Without a
doubt, the PowerCore’s MD3 software is the top solution in terms of limiting and
multiband compression. The algorithm comes from the TC System 6000.
UAD-1-Precision Limiter
Universal Audio Precision Limiter (UAD-1: optional)
From a technical point of view, the Precision Limiter does not even produce overs
with 36 dB overdrive. On the sound quality level it holds up well, but tends towards transient distortion earlier than the brickwall limiter in the MD3 software.
The UAD-1 limiter is a technically reliable and easy-to-use device with very attractive metering. You can certainly do more than just cap peaks with this limiter,
but you should beware of any changes to the sound.
Sonnox Oxford Limiter
Sonnox Oxford Limiter
(PowerCore: optional and native)
The developers at Sonnox have succeeded in creating a very special limiter with
the Oxford Limiter. Here are two particularities, which no other limiter (that I
know of) offers: optical monitoring of overs, which can be expected to occur on
consumer devices; and real-time restoration of overdrive artifacts. Along with
purely technical limiting (i.e., professional level control), you can also use the
Sonnox Limiter for controlled loudness enhancing; it is an ideal tool for the summing bus and as the last plug-in in the mastering process.
Brickwall Limiter of the MD3 bundle
Nomad Brick Wall Limiter – no frills
Nomad Brick Wall Limiter BW2S (native)
The Nomad is the only candidate that quits at 9 dB overdrive and still lets overs
through. At this level, very unpleasant distortion is clearly audible, which are
much weaker with the UAD-1 limiter. Under the same conditions, the TC limiter
still sounds natural and distortion-free. This makes the Nomad only useful for
light top end limiting. The absence of level displays and parameters does not permit any control in the operating range. An output level control and a level reduction display are the minimum elements that a serious limiter should have.
As mentioned in the beginning, the next four plug-ins are designed for mastering. Since many people are also tempted to use these loudness maximizers in the
mix, I will mention the most important devices in the category. Please do yourself
and the people later involved in mastering your production a favor, and avoid
using these plug-ins in the master bus. Mixing is about concentrating on the skillful arrangement of events in the three dimensions. The delicate adjustment of
tolerable loudness in a mix requires another kind of listening. In addition, final
mixes that are too loud are hard to work with. If a lot of reverb is used in a mix,
then it can make sense to use a loudness maximizer, which can be dropped into
the master in order to get the right amount of reverb in the mix. Compressing
during mastering allows the quieter room reverb sounds come forward. For this
you should use a loudness maximizer to set the estimated target loudness of your
production and then check the amount of reverb present. After checking and
adjusting, switch the loudness maximizer to bypass or remove it entirely.
Sonnox Oxford Inflator
Sonnox Oxford Inflator
(PowerCore: optional, native)
The Oxford Inflator from Sonnox is more than just a loudness maximizer – enjoy
it with care! It tempts you to make a recording louder than would normally be
appropriate. The English manage to create a huge variety of sound manipulation
possibilities with just a few parameters. Manual reading and thorough experimentation are both recommended activities in order to comprehend all of the
possibilities for using this device. The Inflator differs from other loudness maximizers that attempt to achieve a great amount of loudness with the least possible
amount of sound coloration. This is why the best use of the Inflator is with drum
groups in order to give them punch, or guitar groups for a more aggressive sound.
All in all, the Inflator is a very warm and analog-sounding loudness maximizer
that has a pleasant effect on the harmonic and transparent aspects of the whole
sound. For users who generally avoid manuals, please note that the pictured parameter values (Effect 100% and Curve 0) represent the setting for the most neutral sound quality. In order to bring the plug-in into the “hot” operating range,
the input level needs to be brought to the highest possible level without overs.
Steinberg Loudness Maximizer shows that the L2 tends to create artifacts, distortion, and changes in stage depth earlier than the Steinberg device. This is why you
need to be careful with dosage.
Waves L2
Waves L2 (native)
The Waves L2 belongs to the most widely-used loudness maximizers and is
quite intelligently laid out. Between the two main knobs there is a link switch
that makes both threshold and output simultaneously reducible. Since the L2 is
level precise – in other words, 3 dB peak reduction is automatically made up for
by 3 dB in the output – you save yourself the complication of doing a loudnesscorrected A/B comparison (this does not apply with very strong limiting). The
output ceiling fader works pretty much like a leveler; with increased loudness
through lower thresholds, level is reduced by the same proportion. This lets you
easily find the limits of good taste, since when bypassed, the original sounds just
as loud as when processed and you won’t be fooled by a difference in loudness
when you do an A/B comparison. After you have carefully adjusted the tolerable
loudness, you can set the output ceiling at an appropriate value. The L2 has an
advantage over Steinberg’s Loudness Maximizer: the ability to reliably adjust output levels with the “output ceiling” parameter, so that you do not need to resort to
leveling or brickwall limiting later on, as long as you have left 0.3 dB headroom
(see Mastering book). When using the L2 as an insert effect in a mix, don’t forget
to turn off the L2’s dithering, so that the signal does not continue on in the 24-bit
resolution format.
The L2 also functions as a pure limiter, which does not affect the lower dynamic levels at all (provided that you do not exaggerate). Blind testing against the
ME Loudness Maximizer
Steinberg Loudness Maximizer (native)
The ME (Mastering Edition) Loudness Maximizer from Steinberg / Spectral
Design is very easy to use and gives the user a quick idea of the amount of loudness available. The “Possible Gain” display shows the amount of loudness that
can be obtained in decibels. When all parameters are in their default position,
this display is a sure indication for a sound, which would not be degraded, when
the LED is still green in the loudest passages. This way, you can see at a glance if
the loudness to be gained is acceptable and you can increase loudness without
negatively changing sound, depth, and spatial information. The other parameters should be used with more care and require careful acoustic monitoring. The
amount of “Possible Gain” can be increased with “More Density,’’ but the resulting
processing can quickly make the Loudness Maximizer begin to audibly change
the sound quality. “Boost” also allows more loudness and makes the signal harder
and more transient-rich. The toggle switch lets you choose between a harder or
softer sound character – you should experiment with this. Compared to the L2 or
L3, the Loudness Maximizer tends toward undesirable changes in sound quality
a bit later. Unfortunately there is no adjustable “Output Ceiling” on the Loudness
Maximizer. Otherwise it can be used as a reliable limiter, safely limiting to minus
0.0 dB. In the mix it can be used in the master bus for checking the amount of reverb while temporarily increasing the loudness, or for moderately being applied
to the drum group track. Unfortunately, the Loudness Maximizer is no longer
for sale officially; with a little luck you might be able to find a “second hand” version. It used to be sold in the Mastering Edition bundle and is the predecessor to
a plug-in of the same name, which is now included in the mastering workstation
Universal Audio Precision Maximizer
Sonnox Transient Modulator
Universal Audio Precision Maximizer
(UAD-1: optional)
The Universal Audio Precision Maximizer is the newest member of among
the maximizers presented here. Its main goal is not sound-neutral loudness
maximizing, but simultaneous sound shaping with tube saturation, similar to
the Inflator. The 50% shape setting results in a “neutral” output in single-band
mode, which nevertheless has a clearly tube character. Processing emphasizes
transients as long you do not use it too strongly. When used it in the master bus,
the Precision Maximizer can be categorized along with the “dangerous” plugins that need constant loudness-corrected A/B comparisons while adjusting.
The mix function make it possible to create interesting effects, going all the way
to tube distortion. Using it with group tracks could be particularly useful for
rough-sounding mixes.
Sonnox Oxford Transient Modulator
(PowerCore: optional, native)
The Sonnox Transient Modulator and the UAD-1 Transient Designer (SPL) are
the only plug-ins in this category I know of that can be taken seriously. The hardware version of the SPL Transient Designer has achieved a legendary reputation.
The somewhat cryptic labeling of some of the parameters of the Sonnox device
makes it more difficult to use.
The Transient Modulator is used to weaken or – more often – to emphasize transients with relation to their length. This means that you can choose to bring out
or attenuate all short or all long transients. Usually the device is used to emphasize transients to create punch, bite, and percussiveness, and thereby bring more
life into the sound. The device can also have a positive effect on the ambience
information of a recording. Some indications for treatment: dullness, flabbiness,
overcompression, etc.
Working with the Transient Modulator requires a certain amount of time to get
familiar with it in order to understand all the processing details. Users might not
be familiar with the idea of increasing dynamic range rather than decreasing it,
which is more usual. The Transient Modulator can freshen up drums and percus-
sion tracks as well as all transient-rich sounds.
Keep in mind that transient processing can result in extreme peaks, which sometimes must be eliminated later. For that you need an excellent limiter that won’t
make the freshly pepped up transients dull. The Transient Modulator should be
used later in the mix, after the level relationships are already established, so that
the modulation can be adjusted in accordance to the track level, as well as any
limiting that might be necessary.
SPL Transient Designer (UAD-1: optional)
In the hardware world, the SPL Transient Designer is already a classic in this
special category and is supposed to sound even better on the UAD-1 platform. At
the laboratory and test studio Universal Audio, I was convinced of this. The two
parameters Attack and Sustain make it easy and intuitive to control transients
and are like a refreshing drink during a summer heat wave. This great plug-in
makes the list of must-have hardware devices even shorter!
In the chapter on “Dimension 3” I already mentioned the criteria for good (in
terms of effective) reverb. Here I would like to examine whether – in a working
context – the following reverb devices deliver the psychoacoustical room and
depth information with their presets.
For testing your own choice of reverb units, take a mix that needs a small room
for drums or percussion and a large room for the vocals. Insert all the devices to
be tested in the corresponding effect channels. In order not to overtax the CPU,
switch on one device (per chain) on at a time. Now look for rooms in all the reverb plug-ins, which should be good for the tasks, sound similar, and have similar
parameter value ranges. Using a short, 4 or 8 measure long loop, compare the
perceived spatial information of the various reverb devices by switching them on
and off. Make sure that the reverb return level is the same in all devices in order
to make a fair comparison!
Surprisingly, you can hear considerable differences in stage depth. The layering
exercise can also be used for this test.
The classic convolution device with a huge library: Audioease Altiverb
SPL Transient Designer on the UAD-1
Audioease Altiverb 6
The Altiverb has become a classic among convolution reverb devices and the
XL surround version comes with a huge library containing not only impulse responses of natural rooms and halls, but also of many reverb units as well as a large
post pro library. Here you can find discrete rooms that make an inconspicuous
but effective layering possible. The device features true stereo processing which
has become a standard even in mastering (for true stereo, see SIR2).
the impulse file representation. Here you can make the reverb tail duller as time
goes on. Or you can make the reverb tail dull at the beginning and then become
brighter towards the end, as if the reverb effect quickly approaches the listener.
This is an interesting parameter for experimentation. The amplitude of the impulse file can be shaped as well (red line) and the reverb can be processed with
EQ. The SIR2 creates impressive spatial depth and provides good, small “ambience” rooms for discrete layering.
RoomWorks Hall from Steinberg
Cubase RoomWorks
Faltungshall SIR² from Knufinke
In the first edition of this book I demonstrated my enthusiasm for the W2 reverb
from Wizoo, which unfortunately is no longer available. Now there is a worthy
replacement similar to that of the W2; though the SIR2 is not a mixture of convolution and algorithmic reverb. The SIR2 convolutions reverb plug-in by Christian
Knufinke is a real stereo reverb device with four channels. Both input channels
L/R are convoluted with two (altogether four!) impulse files and then mixed back
together. In the diagram you can see the “true stereo” routing.
True Stereo four channel
Among the large number of plug-ins that I have been able to examine, RoomWorks
by Yvan Grabit and Charly Steinberg stand out. It is pleasing to see that there are
many small rooms included, which sound good and make intelligent layering
possible. However, the good reverb quality has its price in terms of CPU power, which nevertheless can be reduced with an “Efficiency” knob (smaller value
= higher CPU needs). When the export button is on, then the highest possible
quality will be used for the mixdown. The simple interface which is not graphics oriented (for a change!) is also nice to work with. Spatial information is also
transported well with small doses of reverb. RoomWorks can also be used as a
surround reverb unit.
The requirement for enjoying the full quality potential is acquiring
the extra HDIR (high definition impulse response) from Pinguin.
The plug-in offers everything that one would expect from a highquality reverb. Of particular note is an individually adjustable lowpass filter along the time axis of the impulses. In the plug-in illustration you can see drawable lowpass curve as a purple line over
The UAD-1 RealVerb’s big brother: DreamVerb
Universal Audio DreamVerb (UAD-1: optional)
RealVerb-Pro: Standard equipment on the UAD-1-card
Universal Audio RealVerb Pro (UAD-1: included)
The RealVerb Pro is delivered with the UAD-1 card and is particularly good with
large room simulation. The “Ghost Voice” preset is a good starting point for enveloping lead vocals in a thick reverb tail. Small rooms are less convincing and
the reverbed signal seems very indirect and inappropriately far away. The user
interface is very simple.
The DreamVerb plug-in is the RealVerb Pro’s big brother. It lets you build your
virtual room with a choice of 21 room shapes and numerous surface materials. In addition, you can mix the reflection behavior of three different surface
materials together along a continuous scale. Operation is graphically oriented,
intuitive, and logical. Small rooms sound better than the RealVerb Pro, but the
DreamVerb’s favorite job is simulating thick medium-sized and large rooms, as
well as plate effects. This comes through in the mix.
Plate 140 – the legendary EMT plate reverb on the UAD-1 card
Universal Audio Plate 140 (UAD-1: optional)
The Plate 140 is based on the legendary plate reverb EMT 140 made by the
German firm EMT in 1957; this device created reverberation through staggered
metal plates that transmitted the signal via induction. At the time, spring reverb
technology was the usual fare. The software version contains three completely
independent plate reverb devices that can be controlled from one unit, and
are enhanced by a number of modern functions in order to address today’s
operational needs. The wonderful, soft sound of the EMT 140 has already been
emulated in many later reverb devices. Now the “original” is available with its
unusually thick and full sound. The Plate 140 creates credible depth and is a joy
to the ears. I like to use this reverb for backing vocals and guitars.
MegaReverb from TC PowerCore
TC MegaReverb (PowerCore: included)
ClassicVerb from TC PowerCore
TC ClassicVerb (PowerCore: included)
Although the MegaReverb sounds somewhat artificial to my ears, it offers an atypically
large palette of small rooms for drums and percussion that work well in the context of a
mix. The user interface is logical and allows for spontaneous adjustments. The very good
selection of presets makes it easy and fast to get close to the desired reverb sound.
TC has a long tradition of programming reverb algorithms, which explains the
large variety of reverb plug-ins produced by the company. The focus of ClassicVerb
– which is part of the standard plug-ins included on PowerCore cards – is certainly not small rooms, even though “Drum Box” and “Ambience” do a good job
of providing unobtrusive rooms for drums and percussion. The ClassicVerb is
most at home producing natural rooms for acoustic sounds and provides natural-sounding space. Simple and easy to use, it is still a high-class reverb plug-in. It
is good for acoustic guitars and strings.
Digital Vintage Reverb – the EMT the “Space Heater” for DAWs
The NonLin2 is best suited for special effects and sound design.
TC DVR2 Digital Vintage Reverb
(PowerCore: optional)
TC NonLin2 (PowerCore: optional)
Inside of this plug-in is a high-end reverb unit from the System 6000, which emulates the legendary EMT 250 (also known as the “space heater”). As opposed to
Universal Audios retro graphic style, the TC device has a modern design, which
is influenced by the TC remote control unit of the System 6000. The main use of
this device is certainly lead vocals. Along with the vocal sound of the Beatles, the
EMT 250 is a part of the sound of innumerable classics from the history of pop
The name “NonLin2” says it all: it is not linear. Instead of making the usual efforts
to create a natural-sounding reverb, TC has taken a totally different path in order
to make a creative tool for making widely different kinds of artificial reverb effects. Along with gated and reversed reverb effects for snares and percussion, you
can also use it to create unique vocal ambience sounds. The NonLin2 is a fresh
addition to the sound design world and is not destined to be a general-purpose
reverb device.
Waves IR1 convolution reverb
As a hardware unit, the VSS3 is already an industry standard.
TC VSS3 (PowerCore: optional)
Just like the NonLin2 and the DVR2, the VSS3 also comes from the System 6000.
Under the hood are nearly 800 sound-influencing parameters, reduced to a number of accessible control parameters, which is still quite large; the VSS3 is a highly
professional general-purpose reverb unit that is truly top end. The large preset
library is organized scenically and makes this plug-in perfect for post-production. This plug-in is not an emulation of one of the best reverb units in the world
– it IS that reverb unit! Further details are unnecessary here.
Waves IR1 (native & suitable for APA)
The IR1 is a high-class convolution reverb unit that comes with a very large impulse library. The impulses are natural and assert themselves well in the mix. This
processing-intensive plug-in can also be used on the Waves APA DSP units.
The classic Waves Renaissance Reverberator
Waves Renaissance Reverberator (native)
This Waves reverb was one of the first serious reverb devices available as a native plug-in and offers a large choice of presets. The demands of today’s productions are no longer fully satisfied by the device, but it still is a good, basic naturalsounding reverb unit with the usual parameters.
The Nomad BlueVerb has a vintage design.
Nomad BlueVerb DRV-2080
8-Channel Convolution Reverb from Voxengo
The BlueVerb reverb unit sports a nostalgic design and sounds a bit mushy and
boomy. When the reverb return level is reduced and the basses are lower midrange frequencies are thinned out in the return path, then the BlueVerb delivers
clean spatial depth in the context of the whole mix.
Voxengo Pristine Space Convolution Reverb
The Pristine Space reverb also demonstrates that the quality of a convolution unit
is very dependent on the quality of the impulse response samples that are used.
The interface is a little user-unfriendly because space-saving abbreviations have
been used that do not intuitively indicate which function they control. After a bit
of getting used to, the sober graphics become more apparently logical. It is interesting to note that the Pristine Space Hall can be used with the Voxengo Impulse
Modeler, which allows the simple creation of your own impulse responses.
Under the category “other plug-ins” come all send effect devices like delay, chorus,
flanger, and filter effects. There are so many of these that it would go beyond the
scope of this book to describe them, let alone to find common criteria for evaluation. If the large selection of plug-ins in your host application is not enough, then
please refer to specialized magazines, books, and other publications, or on the
Web sites provided in the appendix.
am pleased that you have read this book with interest. Have fun using the tips
and advice in your productions.
All of the suggested ways of working are not always suitable for all users or all
production situations. Find your own working style that might be a mix of your
methods, my methods, and methods of other engineers and producers. Recently
a successful American producer and engineer explained to me that he always
starts his mix with the vocals and never switches them off until the mix is done. I
have described an entirely different way of working, but he is of course also right.
As I say: find YOUR way of working.
I hope that this book helps you in your daily work and I would like to thank you
in advance for comments and suggestions for the next edition. Please use this email address: info@proworkshops.de.
The tutorial DVDs of the same name dealing with mastering and mixing are a
perfect multimedia addition to both books. For more information, go to www.
proworkshops.de or sign up for the newsletter. On the accompanying DVD you
will find a QuickTime trailer in the “Trailer” folder.
I wish you enjoyment, success, plenty of good work, and happy clients!
Friedemann Tischmeyer
Another decisive factor in terms of the quality of results is beyond our control.
Remember that a good production starts with a good composition and that the
arrangement, the quality of the recording, and the musical performance itself all
determine the quality of the entire work. If one day you have the feeling that you
have followed all the rules of your craft while using good equipment and you still
haven’t achieved the desired results, then change something that occurs before
the mix. Keep in mind that a good mix can reveal deficiencies in the arrangement. If the wrong vocals disappear in a mush of sound, you can bring them
right into the ears with a clearly structured and well-layered mix, but you give the
untrained ear a (subjectively wrong) impression of a bad mix. Then it becomes
necessary to extend your mixing capacities into musical ones and correct the arrangement. Don’t underestimate ear training, even though it might seem as if it
has become very unpopular.
You are welcome to sign up for the newsletter mailing list.
Along with my activities as a sound engineer specialized in mastering, I also offer
private and group workshops for individuals and companies in my studio near
Hamburg, Germany. The group courses are limited to 6 participants, so that each
person can take part and really hear what is going on. In the workshops it is also
possible to use examples that could not be included on the accompanying DVD
or tutorial DVD for licensing reasons. It is particularly engaging to combine a
mixing job with a workshop (Mix & Learn), where the first 1-3 tracks of your
album are worked on together. This gives you enough experience and know-how
to mix the rest of the tracks in your own studio after being helped out at the beginning.
Born 1968 in Flensburg, Germany
Trained as a guitarist and bass player; has played internationally with many
groups, including: Broad (New York), Adrian Thommas (USA), Pat Savage
(Canada), etc.
Since 1990 Recording and mixing engineer experience with German and international artists
1995-1999 Consultant for CD manufacturer (CD production quality control)
Founded Art Beat Mastering Studios; since then more than 400
mastering and remastering jobs for national and international artists such as: Steely Dan, Iron Maiden, Eric Burdon, Johnny Cash,
Inga Rumpf, Joja Wendt, Doc Holliday, Pat Savage, etc.
Project Manager of the world’s first Bluebook release “A Puncto”
Since 2002 Proworkshops (cooperation with Steinberg)
Managed the conference and workshop departments of “VST Days”
in Berlin
Since 2003 Freelance trainer (including for Volkswagen, advisory function for
public radio and television broadcasting, S.A.E. Institute)
Music production for the winner of the “Berlin Today Awards” at
the Berlinale Film Festival with the film “Berlin-Beirut”
Construction of a new surround mastering studio near Hamburg,
mastering and mixing projects, book and DVD projects, workshops
I would like to thank all of the designers and programmers who are responsible
for the development of all these wonderful tools; for their friendly assistance
and answers to my questions. A special thanks to my wonderful wife Angela for
proofreading and for the many great suggestions as well as her loving support for
this book, and my many other projects.
Many technical terms have already been explained during the course of this
book. If you do not find specific terms here, please check in the Index.
Adlib / ad libitum
The term “ad libitum“ comes from classical music and indicates the
musician should add his or her own ornamentation and/or ideas to
the melody. In pop music it refers to variations of the actual melody or
improvisation that is added when the refrain is played for the umpteenth
Audio Engineering Society / European Broadcasting Union. AES/EBU
refers to the symmetrical digital signal standard established by this
organization for data transmission via XLR plug and 110 Ohm cables.
It is the professional equivalent of the S/P-DIF standard.
AFL / After Fader Listening
Mixing console term designating listening to the sound after the fader
for verification purposes. All modifications that occur within the
channel (EQ, inserts, and level) can be heard in AFL mode.
Room or environment reverb. Ambience tracks are often recorded
in studios with good-sounding recording rooms. These tracks are
created for capturing room acoustics. Distinctions are made between
closed, mid, and far ambience. Far ambience is often recorded using
boundary layer microphones (PZM). In reverb devices, presets in the
“ambience” category often imitate discrete recording rooms.
In the context of compressors and gates, the length of time from the
beginning of the sound to the time when the VCA (voltage controlled
amplifier) of the device begins to process signals which exceed the
Aux / Aux-Path / Auxiliary
In the section on split consoles there is information on this along with
a block diagram, showing where the signal is tapped along the channel
strip. Aux-sends are used for headphone mixes and for reverb and effect
devices. They can be either pre or post fader.
Mathematical principle for calculating the acoustic properties of a
room or devices (of a “system”) by using responses from impulses or
frequency sweeps. The results can then be processed into any audio file.
Convolution has been used in a number of reverb devices and plug-ins
in recent years.
Central processing unit
Digital Audio Workstation
Amplification or emphasis.
Decibels Full scale
DC offset
The bass (or base) or fundamental part of a sound.
Brickwall limiter
A limiter that calculates internally with oversampling (multiplication
of the sampling rate) in order to safely eliminate so-called “interleaved
sample overs.” A brickwall limiter is the only sure way to eliminate
digital overs in order to prevent artifacts. More information on brickwall
limiters and digital overs can be found in the book “Audio Mastering
with PC Workstations.”
DC offset designates the presence of continuous signal offset from
zero. You can imagine this visually by the shifting of a waveform on
one side or the other of the 0 line. In the following illustration, you can
Comb Filter Effect
The comb filter effect is an acoustic phenomenon that changes the
sound through differences in distance (and therefore time) resulting
from reflections. If your near field speakers are placed on a five-meter
(16´) wide mixing console, the direct signal will be mixed with the signal
reflected off of the surface of the console. This mixing of signals results
in the addition and subtraction of the various frequencies and leads to a
series of boosts and attenuations along the frequency spectrum, creating
a comb-like image when viewed on a frequency analyzer. Reflections
closed to the listening position in the control room or imprudent
microphone placement or bad recording acoustics can be more or less
affected by comb filter effects.
DC offset can be clearly seen in this voice recording – the audio track cannot be
cleanly cut since there are no zero crossings.
see interference which only occurs under the 0 line and is particularly
recognizable in the preroll before the actual desired signal arrives.
As its name implies, DC offset is a low frequency direct current part of
the signal.
Every audio signal that is not cleaned of low frequency interference
with a highpass or lowcut filter can potentially have DC offset.
These undesirable signals can be present in any microphone signal
The result is that basses – which we wish to hear as clearly as possible
– do not sound clear.
Another disadvantage is that it is difficult to find zero crossings in a
signal with DC offset. If you have audio editing software that has the
function “find zero crossings” so that you can quickly find click-free
spots for editing files, your software will probably refuse to cut the file.
This is not a software bug! After removing the offset signal, you’ll be
able to cut the file wherever you wish. You can even deactivate “find
zero crossings.” I advise against this, since fighting a cause is always
better than fighting a symptom. Every good DAW has a “Remove DC
offset” function.
For dynamic processors and other plug-ins which analyze the amplitude
of the incoming signal in order to adjust internal parameters, DC offset
can cause problems by leading to skewed results. The “plus” side of
the waveform always has a value different from the “minus” one. With
analog devices, DC offset can cause clicks when using switches or noise
when turning knobs.
D.I. Box
Direct injection box. This is a small device for converting impedance,
creating balanced signals, and removing grounds from audio signals.
D.I. boxes are used in studios to convert signals from guitars, basses,
or keyboards into symmetric line-level signals suitable for mixing
consoles or preamplifiers. Ground removal eliminates humming which
often occurs on the stage through the use of different electrical circuits
that have different phases.
DC-Offset displayed on the Pinguin Metering
(wind noise, breathers), synthesizer sounds from analog or plug-in
instruments, as well as string sounds from the most costly sampling
libraries. Sometimes the fault lies in faulty electronic circuitry or defect
cables. In the following illustration you can see DC offset from the
vocal track pictured above (between takes) displayed on the Pinguin
analyzer. The low frequency signal cannot even be heard on a high-end
audio monitor system.
The lower the interfering signal the closer it is to 0 Hertz, which
represents direct current (DC).
Effects of DC offset:
DC offset signals do not have a type of tonal or other relevant content
and “mush up” the bass frequencies. Power amplifiers and (sub) woofers
require a great deal of energy to reproduce these unnecessary signals.
Switching function built into inline mixing consoles, which switches
(“flips”) the input assignments between inline and main channels.
Front of house
This is the mixing position for live concerts.
Gate / gated
The short form of noise gate. A gate is a dynamic processor that –
contrary to a compressor – reduces the output level to a predetermined
amount whenever the signal falls below the threshold value. In live
concerts gates are used to reduce crosstalk, as long as there is no desired
signal on a microphone input. When mixing drums that were recorded
live, gates are used to remove the bass drum or hi-hat sound from the
snare sound, for example. Gates can eliminate interfering signals from
amps on guitar recordings. In the studio, gates are not only used to keep
the signal clean, but also for sound design. For example, some gates like
the Drawmer DS 201 can influence transient envelopes and therefore
have an effect on the whole sound.
The doubling or halving of the frequency of a note. Psychoacoustically,
different multiples (octaves) of a frequency is perceived as the same
Repeating rhythm or series of notes
PFL = Pre Fader Listening
The headphone signal path is usually obtained via aux-send in order to
create a headphone mix for recording sessions.
Refers to monitoring the signal “before” the fader (as the signal flows)
on a mixing console. All changes made in the channel strip (gain, EQ,
inserts, and levels) are not heard in PFL mode because the signal is
monitored before the preamplifier. PFL makes it possible to monitor
the quality of an incoming signal before any other processing takes
Amount of reserves (usually in dB) between the the highest signal peak
and 0dBFS.
Insert / Insert effect
Contrary to send effect paths, which can receive signals from a number
of different aux-send sources, inserts are always located at a particular
point in the signal chain. Insert effects are used when a signal is routed
through a series of effects. Typical insert effects are compressors and
Musical Instrument Digital Interface a standard created in 1982
which created a breakthrough in communication between different
electronic musical instruments and which is still the steadfast standard
in sequencing.
Native refers to computing processes that take place on the computer’s
CPU, as opposed to built-in or external DSP (Digital Signal Processing)
cards or chipsets. DSP, which is built into the computer, is referred to as
“half native.”
Noise gate
In the context of reverb, “plate” refers to the metal plate devices that
were once used to create reverb effects. The German manufacturer
EMT created plate reverb units like the EMT 140 or the EMT 250 in
the 1960s. These were lovingly referred to as “space heaters” because of
the way they looked and the fact that they generated considerable heat.
The term “plate” can be found in the names of many presets and preset
categories, referring to the simulation of these reverb units.
Pot / Potentiometer
Resistor knob
PunchIn / PunchOut
With an overdub (part of a recording which is recorded again), “punch
in” refers to the beginning of recording on top of the pre-recorded
material and “punch out” the end of the new recording. DAWs allow
you to automate or assign these functions to a foot pedal.
The time during which a sound continues to be heard, but usually
see Gate
decreases in volume.
Internal interface for computers that allows audio signals to run
between two applications running in parallel. The program Reason by
Propellerheads was the first application to have a simple connection
to Cubase. Since ReWire is an open standard, it is now a standard for
computer-internal audio communication.
Sony/Philips-Digital Interface is the name of a digital stereo data
transmission standard for consumer devices. Transmission occurs at
least with 16-bit resolution, but can also go to 24 bits. Data transfer
occurs over either an unbalanced 75-Ohm co-axial cable and RCA
plugs (also known as IEC 958), or over optical glass fiber cables
Time during which the signal remains audible before attenuating
Transients are very quick and impulse-like electrical or acoustical
signal attacks or beginnings. Transients often provide the acoustic
information necessary for our ear to identify certain natural sounds
such as a bowed string or piano. Particularly percussive sounds are
characterized by short, energy-laden attacks and are therefore rich in
transients. Transient modulation makes it possible to either expand
or compress transient dynamics (SPL Transient Designer / Sonnox
Oxford Transient Modulator). The main use of transient modulation
making transients more fresh, i.e., expanding them. When too strongly
or badly limited, or signals over 0 dB in fixed-point calculation,
the natural waveform of transients is destroyed, making the sound
unnatural beyond these levels. Special interpolation algorithms exist
to prevent this kind of damage by restoring the original waveform.
Along with restoration software tools, the Oxford Limiter also has
such a function.
Please note that the software included on the accompanying DVD is subject to
the terms, conditions, and license agreements of the individual manufacturers.
The Sound Examples and the wave files (.wav) are to be used only for the exercises. Unfortunately, you may not employ these files for any other use (your
projects, for example) because they are protected under copyright law. Using
these files for anything other than in your own personal learning places is in
direct violation of national and international copyright laws.
1) Software & Interesting Manuals
WaveLab Demo
TC PowerCore
Universal Audio
2) Miscellaneous
Bass Sequence
Frequency Chart
Summing Unit Test
Trailer of Tutorial DVDs
3) Sound Examples (16-bit files)
The Sound Examples are all in WaveLab montage and image (name.cue files)
in the “16-bit Sound Examples” file. You can use these in a number of different
u You can record an audio CD from WaveLab by opening the Montage file called
“Image of Examples-CD.mon;
u You can record an audio CD using the “Image of Examples-CD.cue” file by
using a standard CD recording application (Nero, Toast, etc.);
u You can listen to the examples directly from the WaveLab Montage. This
makes the most sense, because it is easier to jump around and compare. There
is a demo version of WaveLab on the DVD (PC only).
Sound Example #01
Vocals a capella with ostinato loop
The chorus was arranged during the mixing process. In order to round off the
sound in the lower frequencies, I have generated a low voice and quickly played a
bass riff in order to come back to the refrain.
Sound Example #02
Playing with special FX
The first intro is built from a Rhodes sound frozen in reverb played backwards
and filtered. The second intro is built from a reversed acoustic guitar and cymbal.
Sound Example #03
Playing with reverb
The delay is manually created by copying and cutting the sound, then placing the
copies at regular intervals after the original. This often leads to faster and more
precise results than fiddling around with tedious delay parameter settings. On
the words “to Berlin...” the lead vocal disappears in a large room and then comes
back with the intimate nearness of an ambience reverb.
Sound Example #04
Playing with reverb
Here the lead vocals step into another atmosphere with a dub-like delay in the
Sound Example #05
Introducing an instrument with automation
The organ track has typical loudness automation, which introduces it loud, then
brings the volume down after a few bars. Because the organ has been consciously
heard at the beginning, a lower loudness level is fully adequate for localizing the
sound afterwards.
Sound Example #06
Creating more room with panning
In the first example the background vocals are not mixed very wide apart and are
not cleanly separate from the lead singing. In the second example the separation
between the two is much better, since the choir is panned more widely.
Sound Example #07
Creating more room with a stereo expander
Here you have the same situation as in Example #06 except that a stereo expander
has been used. First you hear the example without expansion. The second time,
the stereo expander frames the choir around the lead singing, where previously it
tended to make the lead vocals disappear.
Sound Example #08
Working with loops
This example comes from the included loop processing exercise. The bass and
original loop are very imprecise the first time around, with a washed out bass
frequency range. The long bass sounds of the bass drum interfere with the bass
guitar. The edited example is clean, punchy, and clear.
Sound Example #09
Bass EQ
First you hear the unprocessed and the processed bass soloed, then in context.
The processed bass has more bottom range and is easier to localize, while at the
same time giving the mid-range more space for other events.
Sound Example #10
Vocals with and without the 1176
With the 1176, the voices really come out of the box and the breathers are louder.
All extraneous noises are also brought out, of course. It is interesting to note how
the attack parameter of the compressor brings out audible (and visually recognizable) transient at the “I” in the second sentence. When soloed it seems very loud.
However, in the context of the mix, the singing is crispy and has a nice bite.
Sound Example #11
Saving vocals with EQ
The vocals came from a typical makeshift recording situation that can be found in
(Folder for edits specific to Nuendo und Cubase)
Fxbs (example program files for plug-in settings – you can load these files via
the plug-in function “Load Effect”)
Images (folder for files containing waveform graphics, specific to Nuendo und
EQ settings in Example #11
many home studios. This is especially apparent from the unpleasant room acoustics in the mid-range. Fortunately, the producer recorded this very close-mic’d.
Along with a bass emphasis, this creates a larger level difference between the
voice and the room response, making it relatively easy to eliminate the problem
with EQing. In the illustration you can see how the most disturbing frequencies
can be precisely removed, while at the same time the speech comprehensibility as
well as the air band are emphasized.
Sound Example #12
Percussion with and without depth
In the first four measures the percussion events are given subtle reverb, making
them more present in the room. Starting in the fifth measure, the depth disappears and the percussion becomes flat.
Compression Exercises.cpr (Cubase project files)
Compression Exercises.npr (Nuendo project files)
Layering Exercise
Please refer to the description of the exercise in Chapter 6 (Dimension 3)
Audio (audio folder with wave files)
Images (Folder for files containing waveform graphics, specific to Nuendo and
OMF 2.0 (exercise projects can be opened with any host application which is
compatible with the Open Media File format. Use the OMF import function in
your host)
Layering Exercise.cpr (Cubase project file)
Compression Exercises
Layering Exercise.npr (Nuendo project file)
Please refer to the description of the exercise in Chapter 5 (Dimension 2)
Audio (audio folder with the following audio files)
Loop Editing (Tempo 112bpm/Halftime feel)
01 Bass-90bpm.wav
Audio (audio folder with wave files)
02 Acoustic-Guitar-70bpm.wav
03 Snare-Strainer -90bpm.wav
(effect program files)
Images (folder for files containing waveform graphics, specific to Nuendo und
04 Snare-Attack-108bpm.wav
05 Tom-97bpm.wav
OMF 2.0 (exercise projects can be opened with any host application that is compatible with the Open Media File format. Use the OMF import function in your
06 Vox-93bpm.wav
07 Funky-Acoustic-93bpm.wav
OMF 2.0 (exercise projects can be opened with any host application that is compatible with the Open Media File format. Use the OMF import function in your
Loop Editing.cpr (Cubase project file)
Loop Editing.npr (Cubase project file)
The goal of this exercise is creatively processing a good, but mushy-sounding
loop by using the techniques described in the book for making a clear, punchy
How to work with this example using either Nuendo or Cubase: Read the trackrelated information in the notes part of the Inspector top to bottom. Follow the
steps for processing the loop from the track “Original Loop” to “Rendered Snare.”
In the final mix there are four orange-colored 16-bar rendered loop processing
versions which you can use for comparison (tracks 6, 8, 10, and 12). The tracks
“Original Loop”, “BD Gate”, “BD Tune”, “BD Tuned & Short”, “BD Head”, “Loop”,
and “Snare” are all intermediate steps which are rendered in the orange reference tracks. The plug-ins inserted into the intermediate-step tracks are processing suggestions. If you don’t have the plug-ins, use the best tools you have in order
to understand how the processing examples were made (starting in measure 41).
Tracks 18 to 21 are for comparison between the original and the processed version, both with drums and bass in the full mix.
Procedure for OMF user:
interference between bass drum sound and bass, I cut a few individual beats of
the bass drum and tune them manually with the bass. Use the pitch shift function of your DAW for this. On track 4 starting in measure 21 you can hear the
result up to now.
u Now I will quickly remove the snare beats manually and shorten the bass
drum beats even more radically in order to free up the bass-range from long
deep notes that have too much weight and tend to make the bass-range muddy
(the deeper the sounds, the shorter you should cut them). I have made small
adjustments so that the bass drum matches the bass figure better.
u Since the bass drum is still not clear enough, I will resort to insert effects. With
good EQs I support the bottom, reduce the mids, and create more rhythmic
clarity. The Transient Modulator makes it more sparkly and the increased dynamic is gently controlled with a limiter. The result can be heard in track 6.
u Track 7 is a matter of taste. The previous processing cost us some directionality. With the Waldorf Attack I have been able to tweak the bass drum figure
and mixed the slightly filtered result back in. Track 8 contains the result of
these steps.
u On track 9 a copy of the original loop is given some lowcut so that the hi-hat
or cabassa and the snare overtones can be mixed back in. Track 10 contains the
result of this processing.
u Set the song tempo to 112 bpm.
u Mute all tracks.
u Unmute and listen to track “Original Loop”.
u Now listen to track 2 “Original Loop” together with the half-playback tracks
(Percussion, Bass, Guitars, Keyboards, Pad) to get an idea of the track.
u Use the area starting from measure 61 to reproduce the individual steps which
I took starting in measure 21.
u Track 11 starts with another loop copy extracted from the snare. The short
snare impulse controls the big room reverb as well as a delay and is slightly
compressed and given some high boosting with an EQ. In track 12 you can
hear the result.
u The last four tracks are to be used for A/B comparison.
u Mute the “Original Loop” (track 2) and use for A/B comparisons later.
u We want to extract the bass drum from the loop on track 3. Copy the loop
from track 2 onto track 3 and drop in a 6 kHz highcut filter to reduce the hihat. Then process this with a noise gate. Since the snare peak is very high, it
is difficult to completely separate the bass drum from the rest. Starting from
measure 21 on track 3, you can hear the intermediate results. To reduce the
Lost In The Darkness: K&T Max Quanto & Vivien Levien
Publisher: www.schanzen-musikverlag.de
All other songs: contact Tischmeyer Publishing GmbH
Days We Spend Together: performed by the South African singer Zonke on the
album Invitation by Don Laka. With permission of Bokone Music.
high quality sound conductors
Room acoustics:
high quality cable
high quality cable
high quality cable
Room mode calculation
Wavelength calculation
Miscellaneous Information:
Software manufacturer information:
my Internet site
Bob Katz’s information site
Metering for PC: pro Lab 4.5
information on jitter
Metering for PC: PG-AM 4.5
technical information on digital audio
PowerCore plug-ins /Air Monitor Software
general audio news
Oxford plug-ins for PowerCore
UAD-1 card and plug-ins
Bob Clearmountain’s Site
plug-ins like Vintage Warmer
multiband compressors and EQs
Waves plug-ins
high quality native compressors and EQs
inexpensive convolution reverb (and other products)
freeware convolution reverb
Vincent Burel’s high-end plug-ins
WaveLab, Nuendo, Cubase, etc.
http://www.samplitude.com Magix / Samplitude
freeware & shareware reverb & plug-ins
high-end restoration tools & EQs
Ohmboyz: Modulation delay
Hardware Manufacturing Information:
SPL high quality front-end and back-end
UAD-1 card, front-end, Outboard equipment and converters
TC PowerCore, System 6000
Waves APA DSP units
SSL duende 19” FireWire DSP unit
AD/DA converters and audio interfaces
Air Monitors
D.A.I.S. studio routing systems
Audio Mastering with PC-Workstations (English Edition) ISBN: 978 3 981121 73 5
Tischmeyer Publishing GmbH 2008
Tutorial-DVDs (English, German, French, Spanish):
Internal Mixing Vol. I,
160 min
ISBN: 978 3 940058 02 7
Internal Mixing Vol. II
180 min
ISBN: 978 3 940058 03 4
Audio Mastering with PC Workstations Vol.I
180 min
ISBN: 978 3 940058 05 8
Audio Mastering with PC Workstations Vol.II
160 min
ISBN: 978 3 940058 06 5
Audio Mastering with PC Workstations Vol.III
160 min
ISBN: 978 3 940058 07 2
proworkshops complete 5 (all 5 DVDs in neat box)
840 min
ISBN: 978 3 940058 09 6
Tischmeyer Publishing GmbH 2007
Online shop:
North & South America:
Europe, Africa, Asia & Australia:
To inquire about rebates offered to universities and audio schools
please feel free to contact order@tischmeyer.com
1176 LN 101, 222
32-bit floating-point 23, 32, 46, 51, 79, 215
5MC 64
Acoustic Guitar 199
Acoustics (Studio Acoustics) 52
Ad Libitum 102
Adlib 102
AES/EBU 26,51,62
AFL 91
After Fader Listening 91
Air 20 62
Algorithmix Linear Phase PEQ 217
Ambience 150, 183, 187
Analog Mixer 19, 26, 96
Analog Summing 31
APA32 81
APA44-M 81
Apogee 23
Arrangement 127
Aspects 24, 151, 165
Attack 130, 136, 142, 182, 189, 196, 222
Attack Time 135, 142, 185, 189, 196
aud-1 77,102
Audient Sumo 32, 42
Audio Interface, Minimum Requirements 59
Audioease Altiverb 6 245
Audition Bus 86, 92
Automation 63, 103, 112, 130, 166, 173
Aux/Aux-Path/Auxiliary 32, 36, 44, 85, 91, 149,
156, 161
Back-end 58, 62
Backing Vocals/Chorus 177, 194, 197
Bandpass Filter 124, 138, 195
Bass 54, 72, 108, 114, 121, 128, 135, 141, 189
Bass Drum 108, 121, 127, 158, 181
Bass Range 128
Bass Sector 121
Bell Filter 123, 181, 194, 204
Bessel Filter 124
Bit Resolution 32, 46, 50, 130, 215
BLER value 25
Bottom 108, 127, 181, 190
Breather 97, 119, 197
Brickwall 27, 105, 129, 235
Butterworth Filter 124, 205
Cable 57, 60
Cambridge EQ 124, 205
Choosing Speakers 55
Clean Up Tracks 96
Click 97
Comb Filter Effect 61, 122, 155, 194
Common Basic Parameters 131
Compression 101, 130, 141, 174, 182, 188, 191,
196, 199, 222
Compressor 36, 42, 101, 131, 139, 141, 174
Compressor Types 139
Compressors 111
Convolution 149, 162, 187
Convolution Reverb 149, 162, 187, 245, 255, 256
Core Frequency 53, 121, 123, 183, 186
Correlation Meter 64, 72, 108, 111, 113
CPU 32, 40, 48, 60, 76, 83, 101, 148
Cumulative Truncation 33, 45, 48
D.I. Box 189, 200
D/A Converters 33, 55, 58, 62
Dangerous 2-Bus 32, 42
DAT Recorder 23, 26
DAW 19, 60, 83
dBFS 68, 79
dbx 160 225
DC Offset 50, 121, 201
Decision-Making Capabilities 20
De-Essing 195
Delay 24, 140, 155, 160, 196, 200
Delay Rules 155
Digidesign 64
Digital Audio Workstation 19, 60, 83
Digital Signal Processing 76
Digital Speakers 62
Dithering 26, 32, 49, 52, 80
division of labor 22,165
Doubling Tracks 185
Drawmer DS 201 182,225
Driver 59
Drums 108, 121, 127, 157, 181, 186
DSP 76, 203
Duende 81
Dynamic Processing 57, 101, 185, 222, 238
Electric Guitar 200
Electro Mode 137
EMT 140 249
EMT 250 252
EQ 101, 112, 119, 125, 153, 178, 184, 190, 194, 197,
199, 205
EQ Basic Rules 129
EQing 102, 153
Euphonix 5MC 64
External Reverb Devices 149
Fade-out 105, 128, 167, 176, 198
Fader 101, 129, 150, 156, 166, 173
Filter 119, 168
Filter Types 123
Flip 85
Floating-point 25, 32, 46
FOH/Front Of The House 225
Folder Tracks 97
Frequency Bands 121
Frequency Table 120
Front-end 57
FX 86, 159, 162
Gain 174
Gain increase or reduction 123
Gain Reduction 101, 134, 139
Gate/Noise Gate 158, 182, 185, 186
Glossary 263
Goniometer 70
Group Tracks 99
Hardknee 138
HDIR 246
Headphone 55, 62
Headphone Mix 84, 89
Headroom 23, 69, 80, 101, 142
High Definition Impulse Response 246
Highpass Filter 124
Horizontal 107
Hunecke.de 52, 282
Icon console 64
In-line Consoles 85
Insert 24, 34, 36, 44, 90
JBL 61, 62
Keyboard 113, 201
Knee 138
Knufinke, Christian 246
MixDream 43
Mixer, Mixing Desk, Mixing Console 61, 83, 89, 96
Mixing Console Architecture 83
Mixing Goal 23
Mixing Session 165
Monitor Controller/Monitor Matrix 32, 61
Monitor Levels 61
Mono Compatibility 72, 103, 177
MTC 62
Multiband Compressors 139, 222
Mute 23, 100, 165
LA-2A 101, 231
Latency 40, 59, 60, 77
Layering 147, 154
Left-Center-Right 115, 119 125
Level Measurement 129
Limiter 23, 101, 129, 143, 235
Loops 188
Loudness 28, 68, 140
Loudness Maximizer 140, 238
lower mid-range “misery” 102, 122
Lowpass Filter 124
Mackie 31
Magneto 233
Masking Effect 115,148
Mastering 25, 51, 139
MEQ-5 208
Metering 64
MIDI 76, 95
Mid-range Sector 122
Mix Master 25
Mixdown 96, 163
Mute Switch 165
Native 77, 149
Native and DSP-based Reverb Devices 149
Neve 8816 Summing Amp 43
Neve Filter 215
Noise Gate 158, 186
Nomad Blue-Tube Vintage Compressor 232
Nomad BlueVerb DRV-2080 257
Nomad Brick Wall Limiter BW2S 238
Nomad PEQ2A 218
Notch Filter 124
Nuendo Channel EQ 220
Octave 70, 108, 121, 127
Offline-processing 187
Ostinato 274
Over 23, 32, 47, 79, 129, 182
Overshoot 124
Overview of Recommended Plug-ins 203
Panning Automation 112, 177
Panning, never control via groups 103
Panning Rules 112
Panorama 24, 103, 107, 112
Parametric Filter 123
PAS-Metering 65
PCM 46
Peak Level 235
Peak Meter 64, 69, 108
PFL/Pre-fader Listening 91
Phase Cancellation 74, 111, 112
Pinguin Audiometer 65
Pinguin PG-AM 66, 70
Planning Depth Design 157
ReWire 91
Rhythmic Hearing 21
RMS Level 68
RMS/Peak 137
Roland RE 201 Space Echo 161
RoomWorks 247
Routing 67, 83, 89
RTW Peak Meter 130
S/P-DIF 26, 59, 66, 68
Save CPU/Save Resources 40, 101
Setting up Speakers 56
Plate 163, 249
Plug-ins 60, 77, 82, 203
Post-fader 156
Potentiometer 44
Power Amplifier 62
Precision EQ 206
Pre-delay 151
Pre-fader 84, 91, 156
Proximity Effect 198
PSP Vintage Warmer 230
Pulse Code Modulation 46
PunchIn/PunchOut 87
Q/Q-factor/Quality 123, 126
Ratio 134
Recording Session 19
Release 136, 141
Remotes and Controllers 63
Retrospective 19
Reverb 147, 154, 157, 245
Reverb Device 45, 149, 157
Reverb Rules 154
Shelving Filter 124
Side Chain 138
Signal Generator, WaveLab 53
Single-band Compressors 139
SIR2 246
Sketch 107
Sloppy Syllable Endings 198
Smart AV Console 64
Snare 184
Soft-knee 138
Sonalksis SV-315 Mk2 229
Sonalksis SV-517 Mk2 219
Sonnox Oxford Dynamics 225
Sonnox Oxford EQ 214
Sonnox Oxford Inflator 239
Sonnox Oxford Limiter 237
Sonnox Oxford Transient Modulator 243
Space Echo 161
Speakers 55
Special FX 168
Spectrometer 64, 75
Spectrum Analyzer 70
SPL Meter 54
SPL MixDream 43
SPL Monitor Talkback Controller 61
Split Consoles 86
SSA 2B 42
SSL Duende 81
SSL Duende EQ and Dynamics Channel 210, 233
SSL Duende Stereo Bus Compressor 234
Stage Depth/Layering 100, 147, 150
Stage Plan 107
Steinberg Loudness Maximizer 241
Stereo Basis 103
Stereo Expander 34, 43, 103, 111, 158, 177
Stereo Pan Mode 117
Studio Equipment 55
Sub-Aspects 24
Sub-bass Sector 121, 128
Summing 31
Sustain 141, 191
Tape Machine Style 89
TC 2290 160
TC 247C 224
TC ClassicVerb 250
TC DVR2 252
TC Dynamic EQ 212
TC EQSat 211
TC MD3 Brickwall-Limiter 236
TC MegaReverb 251
TC NonLin2 253
TC PowerCore 80
TC System 6000 79
TC VoiceStrip 213
TC VSS3 254
The Test 35
Three Dimensions 24
Threshold 42, 133, 139
Top-end Compression 182
Top-end Limiting 182
Track Structure 98
Tracks, Group 99
Transient 134, 136, 143, 243, 244
Treble Range 122
Truncation 32, 46, 48, 80
Tube-Tech SSA 2B 42
Universal Audio 1176 LN 221
Universal Audio Cambridge EQ 205
Universal Audio CS-1 208
Universal Audio Dreamverb 249
Universal Audio LA-2A 231
Universal Audio Neve 1073 EQ 209
Universal Audio Neve 1081 EQ 209
Universal Audio Neve 33609 223
Universal Audio Plate 140 249
Universal Audio Precision EQ 206
Universal Audio Precision Limiter 236
Universal Audio Precision Maximizer 242
Universal Audio Pultec & Pultec Pro 207
Universal Audio RealVerb Pro 248
Upper Mid-range 109, 122
Upper Treble Range/Air Band 123
Urei 1178 LN 23
VCA 131, 137, 173
Vertical Dimension 119
Visualizing Sound 70
Vocals 192
Voltage Controlled Amplifier 131, 137, 173
Voxengo Impulse Modeler 256
Voxengo Marquis Compressor 228
Voxengo Pristine Space Convolution Reverb 256
VST Instrument Channel 91
Wavelength calculation 53
Waves APA 76
Waves APA 44 149
Waves C1 Compressor/Gate 227
Waves IR1 255
Waves L2 240
Waves Linear Phase EQ 216
Waves Renaissance Compressor 226
Waves Renaissance EQ 215
Waves Renaissance Reverberator 255
Wordclock 26, 59
Workflow Overview 104
Xpander 78