1
Emulation of Ancient Greek Music Using Sound
Synthesis and Historical Notation
Introduction
In recent years several efforts have been recorded in Greece and elsewhere in
reconstructing Ancient Greek Music instruments, both physically and with physical
modeling techniques (Halaris 1992; Tsahalinas 1997; Politis, Vandikas and
Margounakis 2005; Hagel 2007). Moreover, software for Ancient Greek Music (from
this point and forth: AGM), especially for educational purposes, has been designed
by (mostly) Greek researchers, as it will be described in a later section.
The current project is a new contribution to the field of AGM instrumentation,
since it presents a software application (“ARION”) that can be used as an editor,
composer and synthesizer for AGM the same time. Such an electronic instrument
has never been presented before. ARION is the first instrument of its kind. Its main
advantage is that it provides researchers a user interface that alters scales, accents
and pitch assignments helping them experiment with music forms and scales that
have an inherent fuzziness.
The challenge of the project is to be consistent to the source material and create
an AGM composer with scientific accuracy and the same time to produce a
synthesizing instrument with an easy to use interface targeting non-computer
science experts. But, how can you faithfully reproduce ancient music when you had
never heard something like it? The only safe way is to follow the work of experts in
the field and the actual musical scores. But even these are usually incomplete. Also,
the instruments used at the time were very different than modern or even medieval
ones. Moreover, the true Ancient Greek accent is different from the Modern Greek
2
one and from the one used by foreigners today, the so called Erasmian (Devine and
Stephens 1994), so extended research had to be carried out on the vocal reproduction
of the lyrics. For the part of vocal reproduction, techniques for synthesis of the
singing voice have been used, as it is described later in greater detail. The synthesis
of the singing voice is a research area that has evolved over the last 30 years.
Different aspects of this implicative field involve the interdisciplinary area of
musical acoustics, signal processing, linguistics, artificial intelligence, music
perception and cognition, music information retrieval and performance systems
(Georgaki 2004).
As far as resources about AGM are concerned, researchers and pioneers like
West (1992) and Pöhlmann (2001) have managed to collect and organize a very large
amount of documents, some accompanied by actual music scores, and have given a
scientific insight for a music system over 2000 years old. This project takes their
work and tries to make a connection between that music and prevailing modern
Western Music. Two software modules are produced: ARION and ORPHEUS.
ARION provides a unique interface for AGM composition and reproduction.
ARION can accurately reproduce Ancient Greek melodies, using the sound of aulos,
as well as vocal elements. The function of the application is based on the mapping
and conversion of each AGM symbol to the modern Western notation system. The
reverse process (conversion of Western Music to ancient Greek symbolism) is also
feasible. The user can experiment with the various scales, symbols and frequencies
having the total freedom to “imagine” and hear how AGM really was. ARION has
the potential to be synthesis software for professional music researchers.
ORPHEUS is an interactive presentation for demonstrating AGM instruments.
The virtual environment of ORPHEUS allows the experimentation with the use and
the sounds of the modeled ancient instruments. The Ancient Greek Guitar
(“Kithára”), which was the first modeled instrument, can be virtually strummed
3
using the mouse or the keyboard. The auditory result is ancient Greek melodies. The
application, which is accompanied by information about the history of AGM and a
picture gallery relative to the Ancient Greek Instruments, has mainly educational
character. Its main scope is to demonstrate the Ancient Greek Musical Instruments
to the audience.
The applications ORPHEUS and ARION have been presented to vast audiences
during the International Fair of Thessaloniki between September 8-17, 2006. They
were presented in international conferences, in nation-wide radio emissions, in
newspaper and magazine articles.
The next section provides an overview of AGM, while section “Related Work”
presents previous related literature. The following two sections describe in greater
detail the applications ORPHEUS and ARION. Finally, the last section is about the
objectives of future work on the project.
Overview of AGM
In this section we provide an overview of Music in Ancient Greece. Readers
interested in greater detail should refer to (Anderson 1994; Landels 1999; West 1992).
General Characteristics of AGM
A first elementary clue, which is extracted from the research on AGM, is that the
singer possessed the main role on a musical performance. The soloist’s voice was the
basic “instrument” in a performance. The melody came indispensably from singing.
A musical instrument accompanied the sung Greek poetry. Ancient Greek poetry
and tragedy was inseparable from music (Borzacchini and Minnuni 2001). The term
“lyric” stems from the word “lyra”, or lyre.
4
Figure 1. Singers and performers in Ancient Greece. In ancient Greece, the roles of composers and performers intertwined with each
other (see Figure 1). The reason why not so many handwritten scores from this era
exist today, is that performers used to improvise on the musical instrument, while
the soloist was singing the melody, and not read notes from papers. In a time that
papyri, inscriptions, and other written sources were not readily available, people
used to recall by memory vast amounts of information, unthinkable for
contemporary scholars or performers.
In general, the performer followed the singer’s tempo and sound, but he also
tried to achieve heterophony (by improvising). So, the performer was also the
composer at the same time. As it can be easily conceived from the above, the nature
of AGM was not harmonic. Aristides Quintilianus states: “Music is the science of
melody and all elements having to do with melody” (Winnington-Ingram 1932).
This definition of music fits with the monophonic and melodic structure of AGM.
Pitch System
In ancient Greek theory, there are three basic types of genus: the diatonic
(“stretched out”), the chromatic (“colorful”), and the enharmonic (“harmonius”).
The diatonic genus had a characteristic interval of approximately a "tone" at the top
of the tetrachord, then two successive intervals of approximately a "tone" and then a
5
semitone at the bottom, making up a 4/3 "perfect 4th". The chromatic genus had a
characteristic interval of approximately a minor 3rd at the top of the tetrachord, then
2 successive intervals of approximately a semitone at the bottom, making up the 4/3
perfect 4th. Finally, the enharmonic genus had a characteristic interval of
approximately a "major 3rd" at the top of the tetrachord, then 2 successive intervals
of approximately a quarter-tone at the bottom, making up a 4/3 "perfect 4th".
Instead of using ratios, Aristoxenus divided the tetrachord into 30 parts, of
which, in his diatonic syntonon, each tone has 12 parts, each semitone 6 (Barbour
1972). It should be noted here that the tuning systems are described in terms of
ratios of string lengths or pipe lengths (which corresponds to frequency ratios), so
that we don’t know the absolute frequencies (although they could be calculated).
Ancient Greek music theorists, like Archytas, Eratosthenes, Didymus and Ptolemy
propose exact ratios for the intervals of non-diatonic AGM systems, and even
versions of the diatonic with microtonal modifications (Franklin 2005). The
following table contains different tuning ratios of tetrachords for the three genera
according to literature.
Table 1. Tunings of different genera, as described by AGM theorists
Frequency ratios
Pythagorean diatonic
256:243
9:8
9:8
Pythagorean chromatic
256:243
2187:2048
32:27
Didymos chromatic
16:15
25:24
6:5
Eratosthenes chromatic
20:19
19:18
6:5
Ptolemy soft chromatic
28:27
15:14
6:5
Ptolemy intense chromatic
22:21
12:11
7:6
Archytas enharmonic
28:27
36:35
5:4
6
Researchers trained in Western Music, with its diatonicism and tempered scales,
need additional training to understand the chromatic (Barski, 1996; Politis,
Margounakis, and Mokos 2004; Politis and Margounakis 2003) and enharmonic
background of AGM (West 1992). There are several writers, like Otto Gombosi
(1939) that managed to interpret and recognize the microtonal nature of Ancient
Greek Music theory and practice.
Sources
Although sources about Eastern Music, the successor of AGM, are scattered and
not thoroughly indexed as is the case with its counterpart, Western Music, we know
quite a lot about AGM. There are four main types of evidence by which we know
about AGM:
(1) Textual. Although not so many handwritten scores of AGM have been
saved, there are (luckily) abundant sources about AGM theory. Numerous
treatises in Greek, Latin and Arabic have survived which, mingled with the
study of other material, became integrated into the cultures of all Western
peoples, the heirs of Hellenic learning (Harmonia Mundi 1979).
We know a great deal about the rhythms and the tempo of the music, since
these are reflected in the metrical patterns of Greek verse (Pöhlmann and
West 2001). Adequate knowledge has been gathered about the musical
system, that is, how the scales were conceived and the like, since the works of
several Greek musical theorists survive, like those of Aristoxenus and
Archytas, which are dated in 4th century BC (Winnington-Ingram 1932;
Burkert 1972; Barker 1989).
(2) Notational. Over 60 melodies, most of them fragmented, have survived as
stone inscriptions or musical papyri (scraps of papyrus, the ancient
equivalent of paper) containing musical notation. For instance, the song of
7
Seikilos is one of the preserved compositions of AGM. It is engraved into a
grave pillar that was found in 1883 and is dated between 200 BC and 100 AD
(see Figure 2). Pöhlmann and West (2001) have collected and present 61
original fragments of AGM. While it is certainly true that the hearings are
lost recent research has satisfactorily deciphered AGM notation and rhythm.
(3) Organological. We can infer much about the instruments, using as evidence
surviving fragments of ancient instruments (Halaris 1992; Tsahalinas 1997).
Unfortunately the instruments found at excavation sites were not duly
shaped to reproduce music. The hearings and recordings we have (Halaris
1992; Harmonia Mundi 1979; ECCD 1999) come from reconstructed
instruments, bearing an inherent fuzziness and bias from contemporary
construction techniques and materials.
(4) Iconographic. Depictions of musicians and musical events in vase and wall
painting and sculpture provide valuable information about the kinds of
instruments that were preferred and how they were actually played (see
Figure 1). The interested reader could refer to (Anderson 1994; Bundrick
2005).
Figure 2. The song of Seikilos. AGM notation and its transcription to modern western music. 8
Musical Instruments
There are several references about the musical instruments, which were used in
AGM. Some of them namely are: the lyra, the aulos, the kithara, the hydraulis
(Figure 3), the monochordon, the trichordon etc.
The monochordon, the lyra, the kithara and the trichordon constitute some
examples of ancient stringed instruments. The monochordon (or monochord) was a
rectangular sound box of arbitrary length with a single string, which could be
derived by a movable bridge (Rieger 1996). The kithara was a plucked string
instrument and consisted of a square wooden box that extended at one end into
heavy arms (Hagel 2007). Originally it had five strings, but additional strings were
later added to include seven and finally eleven strings. These were stretched from
the sound box across a bridge and up to a crossbar fastened to the arms.
Figure 3. The archaeological finds of a hydraulis (courtesy: the Dion Archaeological Museum) and the reconstructed module by the European Cultural Center of Delphi (ECCD ). Photograph by G. Ventouris. Courtesy: ECCD. 9
The aulos and its variations were a kind of wind of instrument. In this first
version of ARION, the sound of the musical instrument, which accompanies the
Ancient Greek Singer, is an approximation of the sound of aulos, while the ancient
kithara is the first instrument that was modeled in ORPHEUS.
Notation
In AGM scripts, above each line of Greek is notation that looks mostly like Greek
letters, but is in fact vocal musical notation. Interestingly, the Greeks had two
systems of musical notation, which correspond note for note with each other: one for
the vocal and one for the instrumental melody (West 1992). The instrumental system
of notation consists of numerous distinct signs probably derived from an archaic
alphabet, while the vocal system is based on the 24 letters of the Ionian alphabet.
Some of these symbols can be seen in Figure 4.
The whole system covers a little over three octaves. In particular, it contains
notes between Eb-3 and G-6 (West 1992).The symbols form groups of three. The first
symbol (from the left) in each triad represents a “natural” note on a diatonic scale.
The symbol in the middle represents the sharpening of the “natural” note, while the
third symbol represents the flattening of the “natural” note. For example, the first
triad in the instrumental repertory in Figure 4 corresponds (roughly) to the notes E3,
E#3, Eb3, while the second triad to the notes F3, F#3 and Fb3.
Figure 4. Notes for instrumental and vocal performance, chosen from a pool of symbols comprising the instrumental and vocal repertory. 10
Related Work
The research has performed an amazing evolution over the last decades on the
synthesis of the singing voice. The efforts of the scientists were focused on the
differences between the speaking and the singing voice, taking into account the
special characteristics of the singing voice (e.g. precision in frequency and rhythm,
use of vibrato, etc.). There are plenty of works worldwide on singing synthesis
(Carlson, Ternstrom, and Sundberg 1991; Chowning 1981; Cook 1993; Sundberg
2007). A special tribute should be made to IRCAM’s CHANT project (Rodet, Pottard,
and Barrière 1984) . Originally conceived for singing voice synthesis, CHANT
turned out to be well-suited to simulating other instruments, as well as rich for
synthesis in general. A variant of CHANT coping with diphone synthesis (Rodet
and Lefevre 1997) is close to how ARION interprets and performs the lyrics of
composed AGM melodies.
There is also some relative work on Greek Music. Two text-to-speech/singing
projects on Greek Music are IGDIS (Cook, Kamarotos, Diamantopoulos and
Philippis 1993) and AOIDOS (Xydas and Kouroupetoglou 2001). The latter is a
virtual Greek Singer (vocal synthesizer) for analysis and synthesis of Greek Singing.
In recent years several efforts have been recorded in Greece and elsewhere in
reconstructing AGM instruments, both physically and with physical modeling
techniques (Tsahalinas 1997; Politis, Vandikas and Margounakis 2005; Hagel 2007).
The most notable was the reconstruction of the ancient hydraulis by the European
Cultural Centre of Delphi in 1999. A wide range of other instruments has been also
presented in exhibitions and live performances. For example, Halaris (1992) has
reconstructed AGM instruments. He has exhibited them and his ensemble performs
with them. As prototypes for this restoration have been used fragments of AGM
instruments found in excavations or descriptions of them in papyri.
11
ORPHEUS
According to Greek Mythology, Orpheus (son of Apollo and the muse Calliope)
was a poet and musician. After his wife Eurydice‘s death, he went to the
underworld to ask for her return. His beautiful singing and lyre playing put the
guard dog Cerberus to sleep and moved Hades to let Eurydice go.
Arion was a historical figure, a famous musician at the court of Periander, king
of Corinth. It is said that he played the lyre and sang while on a boat, fascinating
some dolphins, one of whom subsequently saved him from drowning.
Overview
The first of the two AGM applications is about the modeling and presentation of
the AGM instruments to the audience. ORPHEUS is a multimedia application,
designed with Macromedia Flash MX, which also incorporates Microsoft Agents
Technology. ORPHEUS has mainly an educational character. You may download
ORPHEUS from the following site: www.seearchweb.net
Arion
Orpheus .
The application provides an interactive Interface, where the Ancient Greek
Instruments are presented. The first modeled instrument (since more instruments
are planned to be added to ORPHEUS in the future), which is the ancient Greek
kithara, is presented here (for more information about kithara see section Musical
Instruments).
The electronic visualization of the ancient guitar is based on information that was
extracted from writings and angiographies, which have survived from that era. The
user is able to “touch” the strings of the guitar (either using the mouse or the
keyboard shortcuts) and create that way Ancient Greek melodies and sounds.
12
Figure 5. The Ancient Greek Guitar (“Kithara”). On the right is visible the “talking parrot”, a Microsoft text-­‐‑to-­‐‑
speech agent that orally instructs on the use of the “kithara.” Kithara Sample Playback
As Figure 5 shows, the modeled kithara is historically the most recent version of
the instrument, having 11 strings. The sounds of the ancient guitar’s strings has been
implemented according to the correspondence of the Ancient Greek symbols to the
modern notes (West 1992), which can be seen in Table 2.
Table 2. Mapping of AGM symbols to modern notes for the Ancient Greek Guitar.
AGM symbol
Modern Note
C D
E
F
G
A
B
C
D
E
F
The users of ORPHEUS get accustomed with the hearing of the kithara, the
arrangement of the AGM symbols for instrumental performance, and the
correspondence of AGM notes with contemporary ones. The strings of the kithara
13
can be triggered in various ways, depending on the performance preferences or the
learning style of the user. As for the timbre of the sound, as basis were used
segments from recordings of reconstructed instruments’ performances. Using DSP
techniques, sound fonts were created. They are sound data, along with instructions
on how to co-articulate these sounds when a chain of notes has to be performed,
each one having variable duration. In other words, a wavetable synthesizer is
formed.
ARION
Overview of purpose and features
ARION provides a unique interface for Ancient Greek Music composition and
reproduction. ARION can accurately reproduce Ancient Greek melodies, using the
sound of aulos, as well as vocal elements. The application runs on Windows XP,
Vista or any newer version running with .NET Framework 1.1 or a latest release
(which can be found in the application CD).
Apart from the obvious functions that are described in the next sections, ARION
provides some more advanced features for the professional researcher. First of all,
there is full support of microtonality, by allowing any possible frequencies for the
notes, since the user is free to edit all the parameters for each note. Also, all the
idioms of the Ancient Greek Language can be used. The option Synchronize copies
the notes from the Instrumental field and pastes them into the Vocal field, or the
opposite. Finally, there is the capability of music or voice isolation in the final WAV
file.
In order to demonstrate the capabilities of the application, ARION comes with
three ancient songs as presets: “The Song of Seikilos”, “Innovation of the Muse” and
“Innovation of Calliope and Apollo”.
14
C-Sound Instrument
The sound of the instruments that the ARION project performs was made with
the use of CSound. As is well known, CSound is an open-source programming
language designed and optimized for the signal processing of music files (Vercoe
1986, Boulanger 2000). The language consists of over 450 opcodes - the operational
codes that the sound designer uses to build “instruments” or patches.
For the instrumental performance of ARION, a new instrument was devised.
Taking into account recordings from reconstructed AGM instruments, along with
existing Csound patches, a new opcode was created for the .orc (orchestra) file that
is used with ARION’s distribution. Aulos, a wind instrument was simulated,
resembling the hearing of a modern flute.
The Csound module is pipelined with ARION’s Graphical User Interface (GUI).
When triggered, the Csound reads a text-based score file (.sco) and renders the sound
of the aulos. However, as we will see, the vocal performance is rendered outside of
Csound by another software module, the Phoneme Modeler. Since the two voices,
that of aulos and that of the male singer, have to be heard synchronously, ARION’s
Csound module undertakes the task of merging the two rendered sound files “on
the fly” into one file. The quality of the final synthetic emulation result proves to be
adequate, comparing to commercial recordings of AGM.
The handling, the instructions and the directives to the Csound processes,
running underneath ARION’s GUI are committed by ARION, not bothering the user
with any Csound intrinsics. As a matter of fact, only a very experienced user may
notice that Csound is used!
It is true that alternate tools could have been used for the performance of virtual
instruments, like the PW and PWGL visual interfaces the Sibelius Academy at
Helsinki has developed to be used along with the ENP notation system (Laurson et
al. 2005). However, AGM has its own musical symbol repertory, completely
15
different from the Common Music Notation paradigm and therefore new interfaces
had to be devised from scratch for the score-writing process.
Figure 6: The articulation procedure and the physical modeling for voice reproduction. Phoneme Modeler
ARION uses 32 synthesized phonemes for the voice production of the Ancient
Greek Singer. The default phoneme configuration of ARION was designed in a
special interface for this purpose: the Phoneme Modeler.
The Phoneme Modeler is a TCL/TK interface for the modeling and processing of
phonemes. It makes use of the Syntmono server, which manipulates SKINI messages
and is part of the Synthesis Toolkit (STK) in C++ (Cook 1993; Scavone and Cook
2004).
This interface simulates the human voice reproduction mechanism.
The excitations produced by the air flow from the lungs via the vocal cords are
the primary sources of articulation (Figure 6). When producing vowels, a strong
16
blow from the lungs serves as the first excitation, while for consonants a noisy
outburst is used.
By shaping the vocal cords, the various vowels and the consonants are produced.
However, the vocal cords are not able to differentiate the subtle variations of similar
consonants, for instance the plosives /p/ and /b/. Therefore, the pharynx, the soft
palate, the oral cavity, the langue and even the teeth or the nasal cavity undertake
the task to filter the airflow accordingly and articulate the desired phoneme.
Although the phonemes are the basic units of speech, they are not the atomic entities
of the server module. The server uses two basic vocal sources, one for consonants
(white noise generator) and one for vowels (Keller, 1995).
Following Gold and Rabiner (1968), the reproduction procedure is described by a
digital oscillator formed by the two zero and pole filter banks used to produce the
resonance and anti-resonance characteristics of the vocal tract. The mathematical
analysis leads to programming modules (Klatt 1980) described in schematic terms
with the blocks seen in Figure 7.
Programmable variable filters
Pulse generator
Filter
F1
Filter 2
F2
Filter 3
F3
Gain control digital filters
Frequency
enrichment
Voice
White noise generator
Pole filter
Zero filter
Figure 7. The diagram of a cascaded formant synthesizer. The top part synthesizes vowels and the
bottom part consonants.
The shaping of the vowels takes place by passing through a filter bank of at least 3
digital filters with the characteristics of the naturally uttered phonemes (Figure 8).
17
dB
F3
F2
F1
20
0
-20
-40
1
2
3
4 kHz
Figure 8: Formant shaping. The cascaded formant synthesizer along with the formant shaping and the filter
banks simulates the actual procedures for voice reproduction, depicted in Figure 6,
justifying the term physical modeling (Cook 1996).
Indeed, the entities that are exchanged in the client-server multimedia
framework of the Syntmono server are the phonemes of the uttered melody.
These messages are created in the Phoneme Modeler application and are sent to
the server, using TCP/IP sockets as a communication protocol. Once communication
Figure 9. The Graphical User Interface of the Phoneme Modeler. 18
is established, the processing of these SKINI messages, which are compatible to
some extend with MIDI messages, results in the audible reproduction of the
phoneme. The Phoneme Modeler GUI we have devised can be seen in Figure 9.
The interface represents the attributes of each phoneme with sliders. The user
may define the central frequency, the bandwidth and the relative position of the 1st,
2nd, 3rd and 4th formant for each of the 32 phonemes. The users accustomed with
the notion of Sundberg’s KTH synthesis of singing modules (Sundberg 2006) will
feel at home by modeling the computer produced voice altering the shapes of the
formants involved.
GUI
The graphical user interface of ARION can be seen in Figure 10.
Figure 10. The Graphical User Interface of ARION. The application consists of three major interfaces: The Symbol Repertory
Interface, the Ancient Greek Music Interface and the Modern Music Interface.
19
The Symbol Repertory is the container of all AGM Symbols used by the
application. It holds the Instrumental and the Vocal symbols. While browsing
through the symbols the user can see as a tool tip the symbols frequency and the
corresponding modern note.
Figure 11. AGM Interfaces: notes for vocal and instrumental melodic scripting along with the lyrics. AGM time
duration symbols
AGM notes
Vocal performance
Instrumental
performance
lyrics
The AGM Drawing Interface consists of three fields, Vocal Symbols,
Instrumental Symbols and Lyrics (Figure 11). The user can either drag’n’drop a
symbol from the Symbol Repertory to the corresponding field or one can use the
Text Tool (which is located in the Toolbar) to change each field.
The Modern Music Interface has two modes, the Vocal Mode and the
Instrumental Mode. The user can interact with only one mode at a time (Figure 12).
By right-clicking on a note the Edit Modern Note Dialog Box is invoked where
the user can modify the note’s duration and frequency shifting it from double flat to
double sharp and in between (Figure 13).
20
Figure 12. Transcription to the Modern Music Interface. Figure 13. Altering AGM note’s pitch and duration. Many notes of AGM cannot be easily related to a counterpart in modern Western
music. Especially in AGM modes like Phrygian and Lydian, a substrate for the
subsequent development of oriental music systems can be detected. Since often no
accurate correspondence can be made, the software gives users the flexibility to
experiment by assigning different tunings to notes. This can help resolve uncertainty
about scales in a trial-and-error manner by hearing the notes and deciding which
tuning sounds most correct.
If the user wishes to change one or more attributes of a single note, he/she could
right click on the note on Symbol Repertory. On the ‘Edit Note Attributes’ window
(Figure 14), he /she can modify the note’s frequency, the double sharp and double
flat thresholds, the octave, the corresponding western note, as well as the predefined
type (sharp / flat) of the note.
21
Figure 14. Edit Note Attributes – the Dialog Box. To keep things simple, if the user inserts the frequency of a note, the double flat
and the double sharp value can be easily calculated using the corresponding button.
The calculation follows the formula:
By clicking OK, the changes are stored permanently in the Association Table and
the user is prompted to update the already inserted notes in his composition.
The Association Table is the table from which the mapping function reads the
data and maps AGM notes to Western Music notes (Figure 15). Each note has the
following properties: ID (the unique index for each note), CharacterID (the code that
defines the note’s symbol), Frequency, Vocal/Instrumental (note’s category), Double
Flat Frequency, Double Sharp Frequency, Type (the predefined type, e.g. sharp), Note
(the correspondent western modern note), Octave, CorrespondingID (the ID code of
the corresponding Vocal or Instrumental note used for synchronization).
It should be mentioned here that the attributes Note and Octave are those that
define the position of the note on staff and do not affect the sound of the note in the
final produced sound file.
22
For more extensive modifications or when changes in the attributes ID,
CorrespondingID, CharacterID or Vocal/Instrumental are necessary, the user may edit
directly the Association Table from menu “Toolsà Edit Association Table”.
.
Figure 15. Editing the Association Table. The ‘Edit Association Table’ window contains the full list of notes and allows the
user to delete, add, or modify notes using the corresponding buttons. Editing the
“Note” column makes the “Frequency” column change to the standard equaltemperament frequency of the new note name. After the changes, there is the ability
to store the Association Table by choosing ‘Save Changes and Close’, while the button
‘Close without saving’ does not store any changes.
Moreover, the Association Table can be extracted in a .aat (Arion Association
Table) file from menu “Advanced Settings”, so as to be able of reusing it at a later
time. Finally, the predefined Association Table of ARION can be restored any time.
23
Figure 16. Adding new notes. Figure 17. Defining the attributes of the file to be produced – the Dialog Box. The last function is used for creating an audio representation of the current
music document. By clicking it, a dialog box about the status of the output file
appears (Figure 17). The user can configure the final audio output by choosing
which musical elements it should contain: instrumental, vocal and lyrics. He can
also define the tempo of the song (the default value is 60). By clicking on “Export”, a
WAV file is created in the current working directory in a non-real time manner and
a message about successful creation appears on the screen. The user can then listen
to the file from any audio player on his/her system. The audio file is produced by
using Csound’s rendering processes along with these of the Syntomono server.
24
Phonemic Parser
One unique characteristic of ARION is that the user can add the lyrics of an
AGM composition in their original format, i.e. in the polytonal system of writing the
ancient Greek language. Of course, the tool also gives the option to write
immediately in the Modern Greek language (see Figure 11).
The polytonal system (with accents and breath symbols) was invented by
Aristophanes- the Byzantine - in about 200 BC, in order to help the foreign students
of the Ancient Greek language read and spell it correctly, since the Ancient Greek
accent was musical and tonal (Spyridis and Efstratiou 1989), which means that the
vowels were pronounced in a very different way from nowadays.
The Help section of ARION contains explicit instructions on how to install a
polytonal Ancient Greek font.
As previously mentioned, ancient Greek pronunciation is quite different from the
modern one, although the letters used are the same. On the web, there are sites, like
that of S. Hagel (http://www.oeaw.ac.at/kal/agp) with sound examples of how
classic texts are thought to be properly pronounced or AGM melodies should be
performed (Hagel 2007).
Accordingly, ARION has a parser module that transforms words to their
phonemic equivalent, using a set of rules. For instance, if the word “ἄ-
-
” is
met, meaning messenger, then the parser would normally transliterate that to /a//g//g//e/-/l//o//s/. However, the rule based parser would locate that the
morpheme [
] has a hidden /n/ implied, and would transcribe the word to /a/-
/n//g//e/-/l//o//s/, explaining amongst others why the English version of the
word is angel. In the same notion, if the word “ἥ-­‐‑
-
”was met, meaning sun, it
would be ascribed to /h//ɛ:/-/l//i/-/o//s/ since its first letter is a long vowel
with a rough breathing. Again, this explains its approximate English variants:
Helios, Helium, etc.
25
Future Work
The next version of ARION will contain a far more improved sound
reproduction, in terms of the singer’s voice. There will be a much better sound
quality, so as the voice to be heard as more ‘natural’ and less digitized. Our efforts
also focus on an even more exact approach of the ancient Greek accent and the
precise ascription of lyrics. Finally, the next version will have important changes in
the user interface, so as the tool to be very functional and satisfy all of its users
needs. The objective here is to go beyond just educational purposes, and create a
very realistic synthesis result.
Part of our future work is also the enhancement of ORPHEUS with more
musical instruments, like lyre, monochordon and trichordon. This task premises the
electronic modeling of instruments and the corresponding sounds each of them
produces.
Acknowledgements
ORPHEUS has been supported by the “HERMES” project (Program “Science
and Technology Week”, Action 4.4.5. “HERMES” 2006, funded by the General
Secretariat for Research and Technology, Greek Ministry for Development), while
ARION is under the auspices of the SEEArchWeb project (“SEEArchWeb – South
Eastern Europe Archaeology Web”: An interactive web-based presentation of
Southeastern European Archeology. A MINERVA-SOCRATES EU funded project
with code no. 110665-CP-1-2003-1-GR-MINERVA-M). Articles, emissions, sound
archives, and multimedia presentations on ORPHEUS and ARION can be found at
the project’s site with URL www.seearchweb.net .
The authors would like to express their gratitude to Professor D. Pantermalis
(the Dion Archaeological Museum & the New Acropolis Museum) for granting
permission to publish the images of the excavation findings of the hydraulis of
Dion. Many thanks go to Professor and Ch. Giallouridis (European Cultural Centre
26
of Delphi)
for providing permission to publish photos of the reconstructed
hydraulis.
References
Anderson, W.D. 1994. Music and Musicians in Ancient Greece, Cornell University Press,
Ithaca.
Barbour, J.M. 1972. Tuning and Temperament: a Historical Survey, Da Capo Press, New York.
Barker, A. 1989. Greek Musical Writings II, Cambridge.
Barski, V. 1996. Chromaticism. Harwood Academic Publishers, Netherlands.
Borzacchini, L., and Minnuni, D. 2001. A Mathematical notebook about ancient Greek music and
mathematics, University of Bari.
Boulanger, R. 2000. The Csound Book. MIT Press, Cambridge, Mass. (with a CD-ROM of data
and applications).
Burkert, W. 1972. Lore and Science in Ancient Pythagoreanism, Cambridge Mass, 1972.
Bundrick, S. D. 2005. Music and Image in Classical Athens, Cambridge University Press, New
York.
Carlson, C., Ternstrom, S. and Sundberg, J. 1991. “A new Digital System for Singing
Synthesis allowing Expressive Control”, Proceedings of the 1991 International Computer
Music Conference, Montreal : International Computer Music Association.
Chowning, J. 1981. “Computer Synthesis of Singing Voice”, Proceedings of the 1981
International Computer Music Conference, La Trobe University, Melbourne: International
Computer Music Association.
Cook, P. 1993. “Spasm, a real-time Vocal Tract Physical Model Controller and Singer -the
companion Software Synthesis System”, Computer Music Journal, 17(1): 30-44.
Cook, P. R., D. Kamarotos, D., Diamantopoulos, T., and Philippis, G. 1993. “IGDIS: A
Modern Greek Text to Speech/Singing Program for the SPASM/Singer Instrument”,
Proceedings of the 1993 International Computer Music Conference, Tokyo : International
Computer Music Association.
Cook, P. 1996. “Singing voice synthesis: History, current work, and future directions”,
Computer Music Journal, 20(3):38-46.
27
Devine, A.M., and Stephens, L.D. 1994. The Prosody of Greek Speech, Oxford University Press,
New York.
Franklin, J. C. 2005. “Hearing Greek Microtones”, in Hagel, S. /Harrauer, Ch. (eds.), Ancient
Greek Music in Performance, Wiener Studien Beiheft 29, Vienna.
Georgaki, A. 2004. “Virtual Voices on Hands: Prominent Applications on the Synthesis and
Control of the Singing Voice”, SMC 2004, Sound and Music Computing 2004, Paris,
France, 20-22 October, 2004.
Gold, B, and Rabiner, L.R. 1968. “Analysis of Digital and Analog Formant Synthesizers”.
IEEE Transactions on Audio Electroacoustics, AU-16:81-94.
Gombosi, O. J. 1939. “New Light on Ancient Greek Music”, International Congress of
Musicology, New York, 1939, p. 168.
Hagel, S. 2007. Ancient Greek Music. http://www.oeaw.ac.at/kal/agm/
Commission for Ancient Literature, Austrian Academy of Sciences.
Halaris, C., Music of Ancient Greece, booklet and CD, 1992.
Harmonia Mundi – Paniagua, G. 1979. Musique de la Grèce Antique, booklet and CD, HMA
1951015, France. Klatt, D. 1980. “Software for a formant synthesizer “, J. Acoust. Soc. Am., 67(3): 972-995.
Keller, E. 1995. Fundamentals of speech synthesis and speech recognition John Wiley and Sons
Ltd. Chichester, UK.
Landels, J. 1999. Music in Ancient Greece and Rome, Routledge, London.
Laurson, M,. Norilo, V., and Kuuskankare M., 2005. “PWGLSynth: A Visual Synthesis
Language for Virtual Instrument Design and Control”, Computer Music Journal, 29(3):2945.
Politis, D., and Margounakis, D. 2003. “Determining the Chromatic Index of Music”,
WEDELMUSIC 2003, 3rd International Conference on Web Delivering of Music, Leeds, 15-17
September 2003, pp. 93-100.
Politis, D., Margounakis, D., and Mokos K. 2004. “Vizualizing the Chromatic Index of
Music”, WEDELMUSIC 2004, 4th International Conference on Web Delivering of Music,
Barcelona, 13-15 September, 2004.
Politis, D., Vandikas, K., and Margounakis, D. 2005. “Notation-Based Ancient Greek Music
28
Synthesis with ARION”, Proceedings of the 2005 International Computer Music Conference,
Barcelona : International Computer Music Association, pp. 475-478.
Pöhlmann, E., and West, M.L. 2001. Documents of Ancient Greek Music, Oxford.
Rieger, M. 1996. “Music before and after Solesmes”, In: STS Working-Papers, Penn State
University.
Rodet, X., Potard, Y., and Barrière, J.B. 1984. The CHANT Project: From Synthesis of the
Singing Voice to Synthesis in General, Computer Music Journal, 8(3): 15-31.
Rodet, X., and Lefevre, A. 1997. "The Diphone program: New features, new synthesis
methods, and experience of musical use," Proceedings of the 1997 International Computer
Music Conference, Thessaloniki: International Computer Music Association, pp. 418-421.
Scavone, G., and Cook, P. 2004.“Synthesis Toolkit in C++ (STK)”, Audio Anecdotes, Volume
2, K. Greenebaum and R. Barzel Eds., A.K. Peters Press.
Spyridis, H.C, and Efstratiou, S.N. 1989. “Computer approach to the music of ancient Greek
speech”, Acustica, 69(5):211-217.
Sundberg, J. 2006. “The KTH synthesis of singing”, Advances in cognitive Psychology, Special
Issue on Music Performance, 2(3) no. 3, 2006:145-162.
Sundberg, J. 2007. “Synthesizing Singing”, SMC’07, 4th Sound and Music Computing
Conference, 11-13 July 2007, Lefkada, Greece, pp. 9-13.
Tsahalinas, K., 1997. “Physical Modeling Simulation of the Ancient Greek Auloi”,
Proceedings of the 1997 International Computer Music Conference, Thessaloniki:
International Computer Music Association, pp. 454-457.
Vercoe, B., 1986. Csound: A Manual for the Audio Processing System and Supporting Programs
with Tutorials. Program documentation. MIT Media Lab.
West, M.L. 1992. Ancient Greek Music, Clarendon Press, Oxford, 1992.
Winnington-Ingram, R.P. 1932. “Aristoxenos and the intervals of Greek music”, The Classical
Quarterly, 26(3-4):195-208.
Xydas, G. and Kouroupetroglou, G. 2001. “The DEMOSTHENES Speech Composer”,
Proceedings of the 4th ISCA Tutorial and Research Workshop on Speech Synthesis, Perthshire,
Scotland, August 29 – September 1, 2001, pp. 167-172.