Proposal Speech Recognition Calculator (SRC) ECE4007 Senior Design Project Section L02, Calculator Team Alfredo Herrera John Holmes Alex Kee Josh Liang Submitted September 17, 2007 Table of Contents 1 INTRODUCTION..................................................................................................... 1 1.1 OBJECTIVE ........................................................................................................... 1 1.2 MOTIVATION........................................................................................................ 1 1.3 BACKGROUND ...................................................................................................... 1 2 PROJECT DESCRIPTION AND GOALS ............................................................ 2 3 TECHNICAL SPECIFICATIONS ......................................................................... 3 4 DESIGN APPROACH AND DETAILS ................................................................. 4 4.1 DESIGN APPROACH .............................................................................................. 5 4.2 CODES AND STANDARDS ...................................................................................... 6 4.3 CONSTRAINTS, ALTERNATIVES, AND TRADEOFFS ................................................ 7 5 SCHEDULE, TASKS, AND MILESTONES ......................................................... 7 6 PROJECT DEMONSTRATION............................................................................. 8 7 MARKETING AND COST ANALYSIS ................................................................ 9 7.1 MARKETING ANALYSIS ........................................................................................ 9 7.2 COST ANALYSIS ................................................................................................... 9 8 SUMMARY ............................................................................................................... 9 9 REFERENCES........................................................................................................ 10 SRC Group (ECE4007L02) ii Executive Summary Contained in this proposal is a description of the design and manufacturing of a speech recognition calculator targeted towards population with disabilities. These users need a tool to aid in mathematical and arithmetic computation that places physical interaction at a minimum. Currently, there exists no calculator that successfully meets the needs of these users. The SRC (Speech Recognition Calculator) features were chosen in order to facilitate the operation for these users. Completion and satisfaction of these features is the ultimate goal of this design. This proposed design will produce a stand alone and intuitive calculator suitable for all disability types. SRC Group (ECE4007L02) iii Speech Recognition Calculator (SRC) 1 Introduction 1.1 Objective People with disabilities need a tool to aid in mathematical and arithmetic computation that places physical interaction at a minimum. The Speech Recognition Calculator (SRC) aims to be an all-in-one solution for people of all disabilities – limited movement, limited vision, and limited hearing. 1.2 Motivation Typically, calculators require heavy physical interaction in the form of pressing buttons that makes use of very precise motor skills. Such skills could be a deterrent for customers with limited movement and limited vision. Currently, the market for disability-oriented calculators has few choices. In addition, existing products fail to adequately fulfill the needs of the population with disabilities – namely the need for hands-free operation. Most existing products focus solely on limited vision customers, yet still rely on physical movement. 1.3 Background To take an example, Orbit Research offers the Orion Talking Calculator for low-vision customers [1]. The model features speech feedback on actions (button presses and SRC Group (ECE4007L02) 1 computation). However, the calculator still requires the need to physically interface with the device. This proposal offers a design that will produce a hands-free, accurate, and intuitive calculator which will satisfy the needs of its users. The design will be centered on a central microprocessor that will handle most of the computation. Several components will interface with this processor to provide input and output capabilities: a microphone, a speech recognition chip, a LCD display, and a speaker. 2 Project Description and Goals Customers with disabilities need a hands-free, speech-recognition device in order to perform basic arithmetic. Currently, the market for disability-oriented calculators offers few options for this clientele, which include users with limited vision, limited hearing, and limited physical movement. The SRC aims to fill this void by providing features that cater to all disabilities, such as: • Limited Vision o LCD screen that is easy to read and can print large characters • Limited hearing o High volume in order to hear the feedback voice • Limited mobility o Hands-free operation • General o Intuitive operation o Clear and audible feedback speech SRC Group (ECE4007L02) 2 Furthermore, the calculator must perform all the functions of a standard calculator and meet the need of a diverse speaking population. In order to meet the need of the customers, the calculator will feature: • Basic arithmetic operations • Bilingual support (Spanish and English) • Speech recognition capable • Speech feedback • Optional remote operation via internet • Speaker independence SRC Group (ECE4007L02) 3 3 Technical Specifications Table 1. Design Specification Aspects of Designs Design Specification Electrical Specifications Voltage Requirement 5V Operating Current Requirement TBD Audio Specifications Microphone Frequency/ dB 20 Hz to 20 kHz (-44dB) Speaker driver 8 ohm, 300mW Display Specifications LCD Display type STN, YG, positive, transflective [2] Accuracy Specification Probability of missed word 1 out 100 words Dimension Specification Weight TBD Size TBD Environmental Specifications Operating Temperature 0°C to 30°C Storage Temperature -10°C to 40°C Network Specification Connectivity RJ-45 Ethernet cable (100Mbps) The voltage requirement is dictated by the analog speech recognition circuitry [3]. Current requirements will be determined once the circuit is built. However, for comparison, Sensory Incorporated’s RSC-464 voice-recognition microprocessor requires 12mA at 3V [4]. SRC Group (ECE4007L02) 4 The calculator will operate with different volumes and strive to achieve speaker independence. To this end, a microphone capable of capturing frequencies in a range from 20Hz to 20 kHz is necessary to capture all pitches of voice from speakers. In addition, it is estimated that a microphone sensitivity of -44 dB is required to have a system gain of -49 dB to pick up all ranges of voice volumes [4]. These specifications ensure both the speak independence and clear speech feedback from the system. In order to be accurate, the SRC will have to achieve one missed word for every 100 spoken words. In addition, a noise tolerance threshold must be determined once the system is operational. The calculator will have the capability for remote connections via the Internet. The connection will be made via an Ethernet cable connection rated at a 100 Mbps. A standard browser interface will be used to connect and interact with the calculator. For proper device functioning, operation must take place within 0°C to 30°C. For the device memory not to be damaged/corrupted, storage temperature should be within the 10°C to 40°C range. 4 Design Approach and Details 4.1 Design Approach Figure 1 shows a block diagram schematic of the entire SRC system. Voice signal input is provided via a microphone, converted by the analog SR (speech recognition) chip into digital signals, and then fed into the microcontroller for actual computation and display. SRC Group (ECE4007L02) 5 Microphone (Input) Analog Speech Recognition Chip Microcontroller LCD Display Text-to-Speech Chip Figure 1. Block diagram showcasing speech recognition calculator. Table 2 provides the parts that have been pre-selected to build the SRC, based on their compatibility. Table 2. Selected Parts for SRC Parts Function HM2007 by Hualon Recognize and analyze the user’s voice Basic Stamp by Parallax Microcontroller chip that compute the arithmetic operation. LCD by Parallax Will display the answer visually 4.2 Codes and Standards Since the product is to be marketed to the general public, the IEEE Code of Ethics is applicable [5]. Thus, all IEEE ethics codes will be followed, with emphasis on those that concern public safety. SRC Group (ECE4007L02) 6 Furthermore, Section 508 of the Department of Justice guidelines [6] establishes certain criteria that must be met by the device when used by the U.S. government (i.e. public schools), a likely primary customer of the SRC. According to Section 508 of the U.S Guide to Disability Rights Laws, no person can be discriminated because of their physical disabilities [6]. Moreover, this law also implies that agencies must have some kind of accommodation to make technology and electronics equally accessible to people with disabilities. The SRC is one step in the right direction to help federal agencies provide people with disabilities equal access and opportunities to perform to their fullest. 4.3 Constraints, Alternatives, and Tradeoffs Initially, the RSC-464 microchip by Sensory Inc. was considered to accomplish both the speech processing and the arithmetic calculations in a single chip. However, it was decided not to use the VR Stamp because it was expensive, heavily software-oriented, and required additional assembling. Another alternative investigated was to build a speech recognition circuit using a Texas Instruments DSP processor. This route was decided against due to its cost, timeconsuming building, and heavy reliance on C programming. 5 Schedule, Tasks, and Milestones Tasks which have been met at this time comprise of: 1- formation of product concept 2- ordering of parts for product construction 3- technical review papers written for vital components SRC Group (ECE4007L02) 7 Tasks which must be taken on from this point forward are specified in the Gantt chart in Figure 2. Figure 2. Gantt chart showing proposed tasks. 5.1 Meeting Schedule Weekly status meetings will be held on Fridays in the Klaus Advanced Computing Building until project completion. Lab hours for product building and testing will be scheduled as appropriate. 6 Project Demonstration In order to demonstrate the capabilities of the calculator, access will be opened to a wide sample of consumers. The consumer group will verify the calculator functions by performing tasks that test each of the fundamental features. First, a simple simulation will SRC Group (ECE4007L02) 8 be performed to test the MCU math accuracy. Second, a simulation will be performed to test speaker independence. Simulated bilingual capabilities will be tested along with the ability to recognize voices belonging to different individuals. Lastly, the speech recognition ability will be tested through direct visual-audio validation between speaker input and LCD display output. The demonstration will also showcase the limits of operation for the calculator. For example, one speaker will be asked to speak at a specified words-per-minute rate in order to verify speech recognition accuracy at high speeds. 7 Marketing and Cost Analysis 7.1 Marketing Analysis The SRC calculator is an all-in-one, disability-friendly calculator that introduces a handsfree, speech recognition and speech feedback system. The calculator can also be used to remotely tutor users. Since production options on the market is currently very limited, the cost of the calculator should not impede market success. The unit cost of an SRC calculator will be around $250. 7.2 Cost Analysis Tables outlining the details of cost analysis may be found in Appendix A. 8 Summary The SRC provides a solution for customers that need to provide equal-access opportunities to population with disabilities. Federal codes such as Section 508 make this SRC Group (ECE4007L02) 9 a requirement for Federal agencies. The SRC meets such codes in a cost-effective manner by providing an all-in-one solution that incorporates limited vision, limited hearing, and limited movement users. Product verification testing at the end of development will incorporate real users to test each of the mentioned groups as well as additional calculator features. 9 References [1] Orion Talking Calculator, Orbit Research, Wilmington, DE, 1999. [2] Parallax Inc. Technical Staff, Parallax Serial LCD, Parallax, Inc., v.2.0, 2006. [3] Images Scientific Instruments Inc., “How to Build a Speech Recognition Circuit,” Images Scientific Instruments Inc. [Online]. Available: http://www.imagesco.com/articles/hm2007/SpeechRecognitionTutorial01.html. [Accessed: Sept. 15, 2007]. [4] VR Stamp With Serial EEPROM Speech Recognition Module Data Sheet, Sensory Inc., Sunnyvale, CA, 2005. [5] Institute of Electrical and Electronics Engineers, Inc., IEEE Code of Ethics. Piscataway NJ: IEEE, 2003. Available: http://www.ieee.org/portal/cms_docs/about/CoE_poster.pdf. [Accessed Sept 15, 2005] [6] United States Department of Justice, “Section 508: The Road to Accessibility,” United States Department of Justice, Section 508 (29 U.S.C. ‘ 794d). [Online]. Available: http://www.section508.gov/index.cfm. [Accessed: Sept. 15, 2007]. SRC Group (ECE4007L02) 10 Appendix A Table A1. SRC PRICE CALCULATIONS SRC PRICE CALCULATIONS Fringe Benefits Overhead Sales Expense 30% of labor 120% of materials, labor & fringe 10% of selling price Table A2. Labor Cost Labor Cost Engineer 1 (100 hrs x $26) Engineer 2 Engineer 3 Engineer 4 Total Labor Cost $2,600 $2,600 $2,600 $2,600 $10,400 Table A3. Development Cost Development Cost Parts Labor Fringe Benefits, % of Labor Subtotal Overhead, % of Matl, Labor & Fringe $231 $10,400 $3,120 $13,751 $16,501 Total $44,003 SRC Group (ECE4007L02) 11 Table A4. Bill of Materials for 1 Unit Bill of Materials for 1 Unit HM2007 8K x 8 SRAM 74LS373 7448 x 2 LCD Display BASIC STAMP II Keypad Protoboard Microphone Wires, Resistors, and Capacitors Parallax Internet Netburner Kit Total $10 $8.85 $0.80 $1.50 $20 $49 $10 $15 $5 $12 $99 $231 Table A5. Determination of Selling Price Determination of Selling Price Based on: Parts Cost Assembly Labor Testing Labor Total Labor Fringe Benefits, % of Labor Subtotal Overhead, % of Matl, Labor & Fringe Subtotal, Input Costs Sales Expense Amortized Development Costs Subtotal, All Costs Profit Selling Price Net Profit SRC Group (ECE4007L02) 100,000 units $231 $30 $10 $40 $12 $283 $340 $120 $29 $0 $149 136 $285 $13,600,000 12