Page  00000553 Proteic voices in the computer music repertory (1972-1997) GEORGAKI Anastasia Ionian University, Music Department Old Fortress, 49100 Corfu,Greece Abstract In this paper we would like to describe and critically appraise some applications of the scientific research on synthesis of the singing voice in the area of musical composition. In this instance, we would like to give an overview of some importants and rare compositions of the computer music repertory which have used synthetic voices and illustrate the - esthetic compositional issues of the composers who have been involved in the area. 1. Introduction Cette voix divine, monstrueuse, menagante, seductrice, miroir el projection de tous lesfantasmes humains se dedouble et se multiplie jusqu'd la foule (Les chants d'amour, G.Grisey). During the last 20 years, the use of voice in electroacoustic music has seen great change. This is not related to the concrete techniques of the 50's (montage, filtrage, reverbing, flanging, speed changing, etc) but pertains to the creation of unusual "quasi-vocal" timbers by various means (like vocal synthesizers or extended processing signal techniques). In this paper we aim to describe the possible dynamic forms of a computer synthesized voice (in our terms protetc voice) in the computer music repertory of the last twenty years, and to investigate the techniques applied by composers. References to some important compositions of the computer music repertory are used as a vehicle for illustrating these issues. 2. Synthetic speech and the American computer aided "sound poetry" The first attempts to use the computer as a tool of generating synthetic voices in the computer music era began with the experiments of Max Mathews and Lochbaum in the '60s (a bicycle built for two) by the use of the program MUSIC I [Kelly and Lochbaum, 1962]. Some years later, in the seventies, a group of composers including Charles Dodge, Paul Lansky, James Moorer, Tracy Petersen, etc. were greatly interested in the use of the LPC analysis-synthesis technique which opened new directions in the use of the synthetic speech and voice in music creation [Dodge, 1985, 1989; Lansky 1989; Moorer, 1978]. We could characterise some of their works as chef d'oeuvres of the synthetic speech computer music repertory, as well as an attempt to continue the lost "tendency" of the Dadaists and the Futurists of the '20s and '30s, in a movement named soundpoetry. Compositions like Speech songs, In celebration, Lions are growing, Digital tantra, Six fantaisies on a poem, The story of our lives etc..[cf, table 1] introduce hybrid vocal timbres (by means of cross synthesis), artificial "melismes", processed and isolated vocal sounds and "speechlike"sounds with a dynamic spectrum. By the use of synthetic speech composers can easily manipulate the vocal parameters for the poetic necessities of the text, then they can attempt to put the synthetic vocal timbre and the text in the same musical context and apply techniques like the interpolation from monophony to polyphony (talking orchestra's effect) using the same sound material, aiming at the expression of a multitude of human emotions through electronic media [Georgaki, 1998b]. 3. The protelc voices in the context of computer music in the '80s In between, various research projects on the synthesis of the singing voice have been carried out all over the world [Mathews et al, 1961; Chowning, 1981; Sundberg, 1989; Rodet et al, 1984, 1985, 1987,1988;Tisato et al, 1991; Cook Perry, 1989, 1993; Georgaki 1998a].. More precisely as scientific research progressed towards an automated "science-fiction" for the construction of a flexible model of the speaking voice, some researchers got particularly interested in the conception of a singing vocal synthesizer. Making the distinction between singing and speaking, a vocal synthesizer could be a useful tool not only in the understanding of the function of the singing voice and the particularities of singing, but it could also be a flexible instrument which could inspire the imagination of composers towards the construction of artificial voices ex nihilo. Some of these models, like Chant [Rodet et al, 1984], FM synthesis [Chowning, 1981], have been widely used by composers. 3.1.The Chant revolution at IRCAM In the mid-seventies, the first attempts to process the voice electronically by the means of the computer were made at Ircam by Jean Claude Risset in his work Inharmonique(1977) and by Michel Decoust in Interphone(1977). They exploited all the possibilities given by computers at this period concerning the intervention in the microstructure of the vocal signal by digital means. The conception of the powerful program Chant, in the early eighties, by Xavier Rodet and the collaboration of some composers like G. Bennet has opened the way to the creation of some important compositions which use synthetic voices at Ircam.This program enabled composers to construct their own vocal "chimeres", and manipulate with precision all the vocal parameters. Many composers have been strongly interested in using the Chant program in order ot create their own proteic ICMC Proceedings 1999 -553 -

Page  00000554 voices. Conrad Cummings in his Beast songs (1979) created strange sounds in between vocal and animal cries. Gerald Bennet in Aber die Namen (1979) designed extraordinary synthetic voices and compared them to the human ones by underlining the opposition between the reality and the supernatural. The same composer in Winter(1979) puts in interaction synthetic sounds which change from vocal to percussive in his attempt to investigate the perceptive hierarchy of the timbres (vocal-instrumental). M. Pasquet creates for the needs of Atemcristall (1980) extraordinary feminin and masculin voices of a long duration (without respiration) which sing vowels in micro-intervalls and balance between the pitch-extremes perfectly controlled by the composer. Undoubtely, one of the most important compositions cf this period, which makes broad use of synthetic and processed voices, combine the computer analysissynthesis techniques, subtle imagination, and compositionnal structure is the "chef d'oeuvre" of the Computer music repertory Mortuos Plango-vivos voco(1980) by Jonathan Harvey; in this context the composer uses the synthetic vocal material as a vehicule to express his philosopical concerns between the real world (vivos voco) and the metaphysical one (mortuos plango) [Georgaki, 1998a]. In Chriode (1981), which is one of the rare compositions in abstracto of the synthetic voice repertory, the composer Jean-Baptiste Barriire introduces the notion of morphogenesis on the vocal material by experimenting on the perceptual characterisics of the voice and its constraints and attempting a metaphor in the recognition of the forms. In time, many other composers used Chant not only to create strange vocal timbers but also to experiment with its perceptive attributes in collaboration with the psychocaousticians [MacAdams, Saariaho, 1985]. Another composition, Les Chants d'amour by Gerard Grisey left ambiguous impressions concerning the use of synthetic voice in a contemporary musical creation. Here the synthetic voice plays an important role (it interacts to a real choir while the techniques applied by the composer give the impression of an amalgame between the virtual and real timbres). The apotheosis of using Chant in musical creation is the construction of mythological superhuman voices, like that of Apollo, for the opera the mask of Orfeus(1984). In this case, Barry Anderson, constructed a superhuman voice which can pronounce the orders in her own way "sing, sing, sing". 3.2. L'autreface of Phone searching for Anima The characteristic names which compose the below title of this paragraph, are borrowed from works which are going to be mentionned. They refer to the use of synthetic voices in other centers of research (outside of Ircam) and express the esthetic orientation of the composers (Anima, Vision, Phone, I'autre face...). One of the most important compositions of the Proteic voice computer music repertory is Phone (1981) by John Chowning, whose sound material is entirely constructed by the Fm synthesis model; the synthetic voices and their transformation to bell sonorities, or the extension of the vocal boundaries by means of synthesis are paradigmatic [Chowning 1981; Georgaki, 1998a]. Jean Claude Risset created L'autre face in 1983 (for soprano and tape, LMA Marseille) where he uses simulated soprano tones by additive synthesis in order to achieve a dialog between the real voice of the singer and the voice of beyond. Richard Karpen, produced The vision in Italy (C.S.C., Padova) by exploiting the vocal sound data with temporal variations in the creation of a broad scale of dynamic vocal timbres. He experimented with the relationship between text and music and the synthetic voices which resulted represented the confused memory of the writer. in Stockholm (EMS) Lars Gunnar Bodin, created Anima, using the program Chant. He sought to symbolise the fight between a human being and its soul, in a psycho-philosophical context. All these tendencies, at Ircam and all over Europe, which focus on unusal synthetic vocal timbers are indicative of the composers esthetic orientation. They attempt to project their philosophical compositional concerns through the dematerialisation and desimbodiment of the voice by changing the articulation, manipulating with virtuosity the vocal parameters (formants, vibrato, etc..) and using techniques like interpolation, hybridisation, extrapolation etc.. In the majority of cases we have studied, the composers use synthetic voice in order to express imaginary worlds, the beyond, the unconscient and situations that cannot be described by a real voice. 4. The new aspects of proteic voices: processing and resynthesis With the advent of fast and even real-time digital sound production comes the need for elaborated methods who would put these activites under more sophisticated levels of control. In the 90's the composers can create any sort of synthetic voices by processing a real one, having in his dispositions advanced tools which allow them to work on the microstructure of the sound. Let's take the example of Ritual melodies(1990) by Jonathan Harvey composed at Ircam with the aid of the assistant Jan Vandeheede [Vandeheede, Harvey, 1990]. Here the composer works on a rich material of ethnic voices and instruments using advanced techniques of vocal processing supplied by the program Chant (hybridisation between the sakuhachi-voice, and perfect simulation of the Tibetian singing) thus blurring the distinciton between processsing and synthesis. More precisely, we have seen changes in the attitudes of the composers using Chant for the synthesis of vocal timbers at Ircam in the 90s [Georgaki, 1998a].The evolution of the signal processing techniques which approach the analysis-synthesis methods,and the increasing capacity of the computers to process and synthesise sound in real-time (like the new algorithms applied in interactive music) allow to the composer to rethink the synthetic voice in a new environnment. In this way the composer of the 90's is in search of new directions in the synthesis of the voice, by elaborating the vocal material by specific tools of analysisresynthesis ( like Audiosculpt, or Diphone, produced at Ircam) or by processing the formants of a human voice - 554 - ICMC Proceedings 1999

Page  00000555 COMPOSERS WORKS INSTITUTE SOUND ESTHETIC Program./Synthesis TECHNICS ORIENTATION Technique ________ Dodge Charles Speech songs Bell Laboratoiries. 'Pitched speech ", Emergence ofthe (1972-73) Speech synthesis by vocal glissando, computer aided formants and synthetic choir. 'sound poetry". __________________________ analysis byLPC Paul Lansk~y Six Fanraisies on a Princeton Synthesis of choral Auditive confusion poem (1979) University. tones, between the Analysis-synthesis choir-like texvture. "speaker" and the by LPC. "acmanist.' Dodge Charles Any ressemblence is C.C.M./Brooklyn Chorusing, Musical parody, purely coincidental College, Flanging, Forruant Ressurection of the (1980) City University. shifting, doubling., voice of E. Caruso. Analysis-synthesis 'by LPC. ______________ Chowning John PhonB (1981) CCRMAIIRCAM tterpolation betweenl The very important tthe synthetic bell role of the vibrato FM and voices, in the perception of Synthesis/Music 10 Simulation of the the vocal sounds. asso profondissimho. Harvey Jonathan A'fortuos plan go- LRCAMII. Hybridisation, The juxtaposition vivos voco (19381) Chandl Music 10 abstraction, between the real and interpolation unreal world. between real voices (voice-bell) and bells. Barrire J.-B. Chrtode (1981) LRCAM. Extrapolation, Hypervoice with an Chan dFormes interpolation, endless respiration Synthesis in hybridisation, and continuous abst-racto. abstraction. timbral changements Pasquet Yves Marie Atemcrzstall (1981) IRCAM Extrapolation and Dialog between the Chant extension of the artificial. and human Concrete techniques human voice voice. and voice synthesis Artificial vocal crycrescendo Saariaho Kajia Ve~rs le blanc IRCAM Perception Synthetic Vocalises (1982) Chant ambiguity between l/ardin secret the vocal aind non(1985) vocal sounds. Risset Jean-Claude L'aurre ftce (1983) U.M.L-L.M.A.- Simulation, Dialog between the CNRS, Marseille extrapolation, real and unreal. Additive voice fjrom beyond. synthesis/Music V Grisey Grisey Les Chants d'amour IRCAM Extrapolation, Mosaic between (1984) Chant interpolation, real and synthetic hvbridisation voices. Birtwistlc H.arisson Extrait do l'opera: IRCAM Extrapolation Superhuman voice Le masque Chant/Forrnes of the God Apollo d' Orphee (1984) Synthesis in A bstracto Lars-Cunnar Bodin Anima (1984) EMS Stock~holm Hybridisation and The voice of the Chand/Music 10, exvtrapolation subconscious. Synthesis in abstractfo Richard Karpen The vision (1985) C.S.C., Paidova, Construction of The voice of the [CMS/ Music 360, scales of dynamic confursed memory Analysis-synthesis tiubral families by LPC Hiarvey Jonatahan Ritual melodies IRCAM Modelisation of The (1989-90)) Chant Exctra-European intrumentalisation singing techniques of~ voice. and instnrumnets. Timbral transition betweetn the vowoels. Philippe Manoury En Echo (1994) IRCAM Resynthesis and Voices that express t ~processing of vocal the imagination of a MAX/ SIM sounds in real rime girl (f~ormant and pitch ______ _ _ _ _ j _ ___ j fl ower) _ _ _ Table I Prote~ic voices in the computer music repertory ICMLC Proceedings 1999-55 - - SSS -

Page  00000556 in real time (Max/SIM). For example, Tristan Murail in his work L'esprit des Dunes (1994) used the analysisresynthesis of two vocal models (Mongolian singing et Tibetian singing) as a starting point, in order to create new abstract sounds rich in dynamic evolution which preserve the vital caracter of the voice. On the other hand Philippe Manoury presented in his piece En echo 1994) new techniques of analysis -resynthesis of the voice in real-time by the aid of a special algorithm of FFT analysis [Puckette, 1994], included in the MAX environnement which allows the pitch and formant tracking of a real voice and its processing- resynthesis in real-time, giving a new dimension to the electronic cyberspace performance of the voice. The evolution of computer music systems, allows to the contemporary composer to make choices: create in abstracto synthetic voices, analyse and process real voices or process real voices directly by specific tools. From the other side composers who want to create synthetic voices by new systems can choose between the new tools of computer music performance where they can make use of Multimedia apllications, automate nets, and new interfaces [Sawada, Onoe, Hashimoto, 1997;Grubb, Darmeberg, 1997]. 5.Conclusions According to the the references we have made to some important compositions of the computer music repertory we conlude that there are three different types of synthesized voices used in the musical creation (1972 -1997): a) voice in abstracto, where the vocal material is entirely produced by the computer by the aid of a program synthesis (cf. Phone, Chriode) and is restricted in timbral transformations of vowels by the techniques of simulation, hybridisation, interpolation or extrapolation. b) analysis-resynthesis of concrete pre-recorded voice (cf. Mortuos-plango, vivos-voco) which allows the processing and resynthesis not only of the vowels but also of entire sung phrases c) processing of the voice by formant synthesis in realtime (interactive systems). The resynthesis of the voice occurs in real time by applying simple techniques like the processing of the formants, and opens the directions to electroacoustic performance by the computer. in order to justify our beliefs concerning the classification of the proteic voices (being referred in some rare compositions) and the esthetic orientation of the composers, as well as the models and techniques which have been used, we present a table with all these examples[cf, table 1]. REFERENCES Barriere Jean-Baptiste(1983) Chriode I. Proceedings ICMC, New-York. Bennet Gerald(1981). Singing synthesis in electronic music. In Research Aspects of singing, The Royal Swedish Academy of music, Stockholm. Chowning John (1989) Frequency Modulation Synthesis of the singing voice, Current directions of Computer music research, MIT Press, p. 57. Doati R.,Patela Gorresan D.(1985) La materia 6 sorda. La Musica n. 4, p.25 Dodge Charles (1985) In Celebration: the composition and its realisation in synthetic speech. Composers and the computer(44-74).Curttis Roads, editor. William Kaufian, California. Dodge Charles (1989) On Speech songs. Current directions on computer music research(9-17). Max Mathews and J.Pierce ed., MIT Press. Georgaki Anastasia (1998a) Problemes techniques er enjeux estheriques de la voix de sythese dans la recherche et creation musicales. PhD dissertation. EHESS/IRCAM, Paris. Georgaki Anastasia (1998b) Synthesis of the singing voice: links betwen research and creation. In the proceedings of the First symposium on music and computers, Ionian University, Corfu. Georgaki Anastasia (1999) Penser la voix de synthese aujourd'hui. Proceedings of VIth J11M99, Issy-les Moulineaux, Paris. Grisey G., Barriere J.B., Baisn6e P.F (1987) Les Chants d'amour. ICMC'87 Proceedings, Illinois Grubb L., Dannenberg R. (1997) A stochastic model of tracking a vocal performer. In Proceedings ICMC'97, Thessaloniki. Harvey Jonathan (1981) Mortuos Plango, Vivos Voco, A realisation at [RCAM,.Computer Music Journal 5:22-24, MIT press. Jones David Evan (1990) Speech extrapolated. Perspectives ofnew music(28, 1). Kelly J., and Lochbaum C. (1962). Speech synthesis.Proceedings of the Fourth International Congress on Acoustics. Paper G42:1-4. SLansky Paul (1989) Compositional applications of linear predictive coding.Current directions on computer musi cresearch(9-17). Max Mathews and J.Pierce ed., MIT Press. Lorrain Denis(1978) Inharmoniques, rapport de l'Ircam,.p.5. MacAdams St., Saariaho K.(1985) Qualites et fonction du timbre musical,. Proceedings ICMC1985, Vancouver. Manoury Philippe (1991) Les limites de La notion du timbre. Le timbre, mitaphore pour la composition. Ed. Jean Baptiste-Barriere, Ircam et Ch. Bourgeois, Paris Mcnabb Michael (1981) Dreamsong: The composition. Computer Music Journal, Vol. 5, No4, Hiver. Moorer James(1978)The use of the linear prediction of speech in Computer Music applications. Rapports de l'Ircam 6/78. Potard Yves, Baisn6 P.-F., J.B. Barribre (1991). Methodologie de synthtse du timbre: I'exemple des modeles de rdsonance Le timbre mitaphore pour la composition, IRCAM, Ch. Bourgeois, Paris. Puckette Miller(1991) Music and speech synthesis using nonlinear distortion and amplitude modulation. Journal of Audio society. Risset J.C.(1992) Composing sounds with computers. Companion to Contemporary musical thought. Routledge, London. Sawada H., Onoe N., Hashimoto S. (1997) Sounds in Hands-A sound Modifier using datagloves and twiddle interface. In Proceedings ICMC97. Thessaloniki. Slawson W.(1969) Speech oriented synthesizer of Computer music. Journal of music theory 13, n. 1., p.94 -127,. Vandenheede Jan, Harvey Jonathan(1990) Ritual melodies. Document Interne, Ircam. - 556 - ICMC Proceedings 1999