(Lttge)-The Aesthetic, Art and Science of a Composition for Radio

Home , Larry Austin, Pauline Oliveros

CD COMPANION: CONTRIBUTOR’S ARTICLE Transmission Two: The Great Excumion (lTTGE)-The Aesthetic, Art and Science of a Composition for Radio

Lamy Austin, Charles Boone and Xavim Sma ABSTRACT

The three authors each ex- plore an aspect of Larry Austin’s 1990 extended composition for radio broadcast. Composer-writer Charles Boone examines the work for its musical and dramatic inter- TT:TGE AS HYBRIDART- ments, computer sounds, pre- est and for its significance as a recorded texts-are being heard hybrid of Horspiel, with elements of A LISTENER’S PEMPECTIVE in the same way by radio theater, portraiture and sound col- lage. Its scope, means, intent, reali- Charles Boone listeners, and wonder especially whether the sound mixes zation and compositional approach are elaborated by Austin, the com- HORSPIEL and balances are the same. One poser. Throughout his section, he can imagine audience mem- interspersesaphorisms drawn from Sonic composition conceived to be heard through loud- bers bringing along portable his conversations with the seven speakers is one of the unique and original art forms of the radios, like spectators at a ball protagonists of his ‘sound movie’. second half of the twentieth century. Concert halls were Austin’s collaborator, computer- game, just to find out what music scientist Xavier Serra, expli- assumed at the beginning to be the proper venues for the outside world is, in fact, cates the technicalfieoretical presenting this kind of work, but it was not long before hearing. A comparison of a re- genesis of TT:TGE‘s computer- composers began searching for other options. Morton Su- cording of the live perform- generated speech sounds. botnick, for instance, was among the first to try subverting ance in Denton, Texas, with a such traditional presentation by composing an electronic recording taken from its radio work (SilverAPglesoftheMoon [ 19671) thatwould be available broadcast showed considerable only on a long-playing phonograph recording [l]. This differences, particularly in the sound levels of the texts. In piece was intended to be played on a normal high-fidelity the radio mix these sound levels were significantly higher system in listeners’ homes and in no other way. than in the concert hall mix. In the past decade another idea for bringing original sonic art directly into the home-again, work intended for only this means of dissemination-has begun to flourish. DIALOGUE AS PORTRAITURE This is Hiirpiel, a sonic art form that has developed into something truly indigenous to radio. Hiirspiel, or ‘listen The question of balance among the various materials, then, play’, shortened from the term Neues Hiirspkl, was coined by is extremely important, and the levels of perceptibility seem German radio producer Klaus Schoening, who, since the to have been a key concern for Austin in assembling the late 1960s,has created radio plays by combining sounds and layers of sound. Of course, the recorded dialogues were words in nontraditional collages. Schoening has since central to this consideration since they are the basic building broadened the concept, now calling it .am acustica. It is, as blocks for the entire piece. Because of the nature of the he says, “for ears and imagination” [2]. dialogues, TT:TGE has a fundamental relationship with As the name suggests, there has been a particular flurry documentary reportage, even though the piece is far re- of such activity in Germany, but there are now composers moved from any traditional work in that genre. and other artists involved with sound and theater who are In the final piece, the words operate on two levels: the working on similar projects all over the world. original, purely informational precorded conversations (what I call ‘texts’) exactly as they were spoken; and the fragments of these same texts that Austin processed at TT:TGE Stanford University’s Center for Computer Research in Music and Acoustics (CCRMA). Obviously, these processed In terms of its forces and duration, Transmission Two: The fragments (Austin’s ‘sonic furies’) were pushed well beyond Great Excursion (?T:T(;E)(1988-1990) may well be the most comprehensibility into the realm of pure sound. Perhaps ambitious Hiirpieelwritten so far. It is somewhat of a hybrid not so obviously, the unaltered texts often function similarly case, however, in that it must have a concert performance, with all the preparation that implies, in order for the radio Larly Austin (composer, writer, teacher), 2109 Woodbrook, Denton, TX 76205, U.S.A. transmission to be possible. This necessity for a concert Charles bone (composer, writer), 2325 Bear Gulch Road, Woodside, CA 94062, U.S.A. performance of something really intended for radio pre- Xavier Serra (acoustician, musician), Center for Computer Research in Music and Acoustics, Dept. of Music, Stanford University, Stanford, CA 94305, USA. sents a certain ambiguity: listeners in the hall might well Received 12 September 1990. wonder if all the sounds they are hearing-voices, instru-

0 1991 ISAST Pergamon Press plc. Printed in Great Britain. 0961-1215/91$3.00+0.00 LEONARD0 MUSIC JOURNAL, Vol. 1, No. 1, pp. 8148,1991 81 in that they are not always above the As this brief description shows, the the chorus’s contrapuntal lines-as of threshold of comprehension; sometimes movement has a clearly developed the growth of the texture and the pro- they are crisp and clear (particularly at form. The combination of musical ele- gression from particularity to general- beginnings of episodes), and at other ments with the prerecorded conversa- ity. Perhaps, the traditional A B A‘ form times they blend into the background tions from which these elements were and the use of strict contrapuntal pro- while other musical events take prece- derived is carefully conceived and cedures somehow reflect the academic dence. In traditional documentary such worked out to heighten the musical setting of CCRMA, where Chowning license with a text would be unthinkable. message as well as to make the ‘picture’ does his work. This, however, is only Why, then, did Austin go to the of Ashley somehow a true one. The speculation. trouble of recording all these conversa- mood and spirit of this movement The Jerry Hunt Episode, which fol- tions and then use them in ways that come directly from Ashley’s own work; lows on the heels of the Chowning, often preclude the possibility of their there is a very relaxed quality, studied provides an abrupt and lively contrast. being fully heard and comprehended? in its casualness, that anyone familiar It is a leap from the structure of aca- When we understand that his intention with his music will recognize imme- demia to the freedom and chaos of the in 7T:TGEgoes well beyond the simple diately. The general effect of the world of rap music. Here, there is the transmission of information, we can be- whole episode is almost like sitting in a driving beat of a drum machine from gin to see this work not so much as a crowded bar with a blues piano playing the first measure to the last, in contrast documentary, didactic effort but rather in the background, snippets of many to the absence of pulse that char- as a series of composer portraits on a conversations drifting in from other acterizes the previous episode. The more abstract level-a musical object booths (sometimes heard and some- choral singing of the Chowning Epi- based on the words, ideas and personas times only perceived), occasional laugh- sode is replaced with choral speaking. of a particular group of composers. ter, and a gentle barrage of group noise The singers are assigned rhythmic pat- and other ambient sounds. This is a terns that match fragments of Hunt’s clever musical portrait that transcends spoken text, which are repeated over THEEPISODES the spoken text that was the starting and over for long stretches of time. The The interaction of the unaltered re- point for its composition. clarity of the John Chowning Episode corded conversations and computer- All of the episodes are similarly ex- gives way here to a wild free-for-all that processed sonic furies with the voices of ecuted, using all of the elements-and mounts to a blast of noise. This steady- the chorus and the improvised instru- especially the choral and instrumental state sonic mass does include some mental music of the score is a major elements-to reflect some of the dis- variety as the movement goes on, but element in these portraits. A good ex- tinctive characteristics of each of essentially, it can be heard as a large ample of this amalgamation of diverse Austin’s subjects. The use of textural wave of sound that rolls along until it components is heard in the RobertAsh- material in 7T:TGE is found in a num- suddenly diminishes at the end, like a ley Episode, ‘The ‘I actually wanted to ber of Austin’s earlier works as well. His phonograph that is switched off while buy’ Blues”. The movement beginswith process, here and elsewhere, is gener- the needle is still in the groove. The text keyboard and percussion music. The ally to provide materials (on tape, for in this movement is almost never heard keyboardist improvises in a slow-blues players, etc.) that he then literally well enough to be understood. None- style reminiscent of music in Ashley’s and/or figuratively ‘switches on’ and theless, it is Hunt’s amusing and lively video opera, Perfect Lives [ 31. (The har- lets run, allowing them to interact and monologue (it can be read in the score) monic structure has been laid out for connect, however serendipitously. A that provides the overall character of each part, and audiotapes of slow-blues comparison of the first two episodes- the music, and the quality of his rapid style music that model the intended the John Chowning and the Jerry banter clearly influenced the way in effect have been provided to each per- Hunt-clarifies this and sheds light on which the choral and instrumental former.) Likewise, the percussionist im- the organization of the other sections parts were conceived. provises a jazz accompaniment using as well. The remaining four episodes also wire brushes on suspended cymbals. The John Chowning Episode is in have their own special formal charac- These musicians are soon joined by the three parts (A B A’), which one can teristics, use of materials and modes of celletto (electric cello) player, who plays easily see by examining the chorus part portraiture. The David Tudor Episode fully notated, blues-inspired licks, and separately. The first and last of these features a tango, in its instrumental and by chorus members, joining in one by parts (A and A’) are essentially vertical vocal parts, that is heard throughout, as one, singing and speaking. By this time, and harmonic in style, while the long well as rain sounds in the electronic recorded dialogue between Ashley and middle part (B) is linear and contra- part that relate to Tudor’s own com- Austin is also playing and continues for puntal. The B section is canonic in position, Rainfme.stZ-N( 1968-1973). For the full duration of the movement. The form, with the imitative voices slowly the most part, the chorus sings, but texture builds and becomes more com- entering in pairs (sopranos,followed by there are whispered sections as well, plex, leading to a high point three quar- altos, then tenors, then basses), thus The Max Mathews Episode is the ters of the way through the piece, where thickening the contrapuntal texture gentlest of the seven, perhaps reflecting all chorus members are singing in uni- and building over a period of about that aspect of the subject’s character. son. The movement ends with a whis 5 min to the piece’s first major high For the largei- part of the movement the pered section of eight independent point. With these voices are layered the men of the chorus whisper, while the chorus parts and continued instrumen- spoken text and the sonic furies, plus women sing. This all begins in fragmen- tal improvisations. The choral materials the instruments, which provide a long, tary fashion, but as the movement pro- complement the computer-music sounds sustained droning sound and rich gresses, the materials become more with which they are layered in this clos- coloration. One is not so much aware coherent and coalesce into the late ing section. of the individual voices-particularly of nineteenth-century song “Daisy Bell”,

82 Au.slin, Boom and .ba,Transmission Two: The Great Excursion ending on a full-blown F-major chord. tants; and prerecorded dialogues. The continuity of TT:TGE are seven epi- In the Morton Subotnick Episode, the dialogues heard throughout the work sodes, each devoted to a dialogue be- chorus begins whispering in unison, present a chronicle of my conversa- tween me and one of my longtime com- which evolves into spoken polyphony tions, recorded between November patriots-adventurers with distinctively near the end. The Pauline Oliveros Epi- 1988 and July 1989, with fellow com- etched musical and technological pro- sode is marked by chordal choral writ- posers Robert Ashley, John Chowning, files. Key segments from more than 11 ing throughout, with the chorus singing Jerry Hunt, Pauline Oliveros, Morton hours of these recorded conversations and sometimes humming as well. Near Subotnik and David Tudor and com- were distilled to 107 min: the Prelude its end the audience is invited to join in puter-music pioneer Max Mathews. and Episodes 1-4 (Chowning, Hunt, by humming-a kind of meditational With me, they reflect on their work as Ashley, Oliveros) form Part I; Episodes participation that comes directly from composers and music experimenters 5-7 (Tudor, Mathews and Subotnik) Oliveros’s practice of involving listeners today and through the last 30 years of and the Postlude form Part 11. The epi- in her own performances. dramatic technological developments sodes range in duration from 12 to 16 in the way music is created and pre- min each. sented. The musical continuity of 7T:TGE CONCLUSION unfolds through the episodes with Austin’s TT:TGE is an assemblage of a choral settings and computer-music very large number of materials, ranging FORM AND CONTINUITY transformations of select aphorisms uttered spontaneously during the con- from traditional instrumental and ve I manipulate convention . . . the versations I recorded-ssences, I feel, cal practices and straight-spoken texts expectations of perjmance and the of the protagonists’ aesthetic and com- to innovative live-performance prac- memoly and the comparisons and positional approaches. The role of the tices and computer-generated sounds. ideas of perfimance and the relation- chorus in my sound movie is choric By using recorded dialogues (more ship to, uh, to the gestures that are as commentator, not unlike its role in clas- normally encountered in a documen- associated with perfmance. . . . I sical Greek theater: musing poetically, tary context) to generate abstract and decided to compose with that, . . ty reflecting philosophically, interjecting purely musical materials, Austin creates getting rid of the idea of music itself assertively, commenting amusedly or a type of historic capsule of some of the and just p’ngdirectly to just mocking ‘rap’-ingly. composers and music of our time, while memMy and retrieoal and histoly of simultaneously interpreting and com- euerything you be done. . . . Then, menting on the materials therein. Per- you can cany, kind of in a way . . . haps this fresh approach to portraiture it’s a little bit like exchanging cas- APHORISMAS SONIC and documentation is more informative settes in a social. . . in a socially MATERIAL: TECHNICAL (as well as being more entertaining) in charged atmosphere. . . . I have no REALIZATION the long run than if Austin had simply feelings about taste or anything. . . . laid out the facts in a more straightfor- He’s like a Mozart, . . all the con- We studied all thosefm in school. ward but less original manner. ventions, he. . . accepts them all. I . . . the^ were the models for us, and mean thqr ’re all there: the clichks we had no contact with them at all. about sex and power and black and . . . There’s no way that I could be MODELING white and monq and everything comfmtable uniting a sonata. . . . I mean that’s exttmely sentimental. . . TGE-THE and the chords and the rhythm me TT: chines. . . . I’ve always cheated. . . . because the language doesn’t contain COMPOSER’S What do you wantfiom me now? It’s any words. . . mechanical marvels, magic. . . mystery machines. . . . I PERSPECTIVE kinda’ like God, you know, is a peteer up there, you know, and we’re feltthisgrowingexatementin me. . . Lamy Austin all playing down below, and this because I hadn’t bought anything. . . machinery is just grinding. . . but I almost like being in love. It was A RADIO PIECE understand it better, as I’m getting almost like sex. I felt the great desire I compose computer music. . . . It older myself because, you can hear to own a new piece of equipment. I was absolutely an ear discovery. . . . that coffin snappin you know! And hadn’t owned anything. I became like the most avid consuw. I Music is as abstract as any mode of you ’re thinkin :“Snap, snap, snap, actually wanted to buy. art. . . . Push the technology, push snap . . . it ’s snafiin ’for me. ,,And the technology! when you hear that snap, you think, -Robert Ashley “Do. . . ha. . . how. . . you know, I -John Chowning want a good snapshot, I want a good snapshot left. The tens of aphorisms selected from The premiere of my composition for the seven conversations created rich radio, Transmission Two: The GreatExcur- -Jerry Hunt material for a wide range-literally sion (TT:TGE), was broadcast live in a hundreds-of sonic transformations. public performance from the Berkeley The first live broadcasts of 7T:TGE cul- Each aphoristic utterance was first ana- campus of the University of California minated 2 years of planning, recording, lyzed, then transformed and resyn- [4].TT:TGEis scored for a chorus of 40 modeling, computing, experimenting thesized with a unique computer-music voices (soprano 1-2, alto 1-2, tenor with and composing the materials for software system called Sound Analysis 1-2, bass 1-2); a computer-music en- what I have come to call my ‘sound System (SANSY), recently developed for semble, including keyboardist, cellist movie’. sound analysis/transformation/resyn- and percussionist, with technical assis- Central to the form and narrative thesis. Named Spectral Modeling Syn-

Austin, Bmeand ha,Transmission Two: The Great Excursion a3 thesis (SMS), it was developed by com- These transformed sonic apho- from listening, exactly in the moment”, puter-music scientist Xavier Serra at rism-what I call my ‘sonic furies’- quietly spoken by Pauline Oliveros) . Stanford University’s Center for Com- are heard as musical events in diverse puter Research in Music and Acoustics timbral/textural combinations, invoked (CCRMA) [5].While this experimental through the piece by three musicians PITCH AND of a computer-music ensemble with system was being perfected and tested DURATIONAL MAPPING in 1989, I was invited by CCRMA Direc- unique computer-music instruments: tor John Chowning and Serra to ex- ‘controllers’ of multiple Sequential . . . exactlyfim listening, exactly plore the compositional potential of Prophet-3000, l6bit digital sound ‘sam- in the moment . . . to experience the SANSY in the creation of sonic materi- plers’ (a specialized type of digital body as a resonator. . . the body as a als for TT:TGE. After an intensive intro recorder) [6]. Through programming resonator of sound. . . . Hear evuery- duction to the complexities of SANSY the protocol of Musical Interface Digi- thing that’s gdng on . . . be open to operation, I undertook a virtual tal Interface (MIDI) interaction between the sounds . . . deep listening. . . marathon of intensive experiments the controllers and 40 banks of over 100 becoming a part of what you ’re listen- during my composer residency there ‘presets’ of sonic events stored on a ing to. . . , That was a nice moment. from May to July of 1989 and again in cartridge hard-disk drive for fast access -Pauline Oliveros December of 1989, when I completed by the multiple samplers, the musicians the final sonic transformations for play and layer my sonic furies through 7T:TGE. I can state that SANSY is one the course of the piece. Typically, the The rhythmic, melodic, harmonic and of the most powerful tools for sonic furies are as short as a compressed 1-sec textural material for much of the choral transformation I have used in my work utterance (e.g.“space to sing by itself”, and ensemble music was extrapolated as a computer-music practitioner over uttered by David Tudor) or as long as a from the durational and harmonic the last 22 years. stretched 32-sec utterance (e.g.“exactly partial data yielded from the Fourier

Pauline Oliveros Eoisode: Madrieale orofondo nella terza orattica Fig. 1. Extrap* lated pitches/ durations from Fourier transform of Pauline 4II - Dnp Ik’ - biq ex- Oliveros’s uttering ‘deep 4 listening‘.

84 Ausfin,Bmnr and Sma, Transmission Two: The Great Excursion Table 1. Vettical phh/durational combinations extrapolated from frequency of vocal Organic growth was the nature of formant. the medium for me. . . . We all pew Syllable: deep lis- thing [sigh...I up under the sign of McLuhan . . . medium and message. . . . It’s the Begin-time: 0 .225’ .35’ .725”-1.175” - gesture. . . . The color is the lan- Hertz=pitch 195=G3 196=G3 219=A3 [noisel guage of the gesture. The gesture is 481 =B4 392=G4 403=G4 [noisel the motive, and the color is the lan- 585=D5 588=D5 638=D#5 [noisel guage, just as words+itches-are 978=B5 782=G5 752=F#5 [noisel the words of language. . . . I have to 1175=D6 981=B5 972=B5 [noise] mean something, and VImean some- 1370=F6 1176=D6 1036=C6 [noise] thing, it will be colarfid. . . , There was a hippopotamus, a purple hip 1761=A6 1582=G6 1562=G6 [noisel popotamus that was the sink in the 1954=B6 1965=B6 1907=A#6 [noisel bathtub, and it occurred to me that thejrst thing I should do un’th a big transform and the subsequent analyses ralist world, one where human rights computer is to turn it into a purple provided in SANSY’s graphic and nu- are freely and creatively experienced hippopotamus. . . . I’m the master of merical representations of harmonic and expressed. With that idealized ref- the medium. . . . Technology has to and amplitude content of the sampled erence, I experience and assimilate all make the meaningfulness more mean- utterances over time [7]. Figure 1 kinds of music. But, in spite of my ap- ingful. . . . rfwe lived in a world illustrates this process in the opening petite for tasting the fruits of diverse where the butterjly was a metaphor, of the Pauline Oliveros Episode. musical expressions, the influence they how would politics be, how would ed- The utterance “deep listening” is have on my own work is elusive and per- ucation be? I wanted it to work stretched 10 times its original length of haps not one that a composer should so globally, so that it could work in- 1.175sec toadurationof 11.75sec.The selfconsciously contemplate. I believe dividually for a piece of music. pitches for the eight-part choral sonori- there is a process of synthesis taking -Morton Subotnick ties were then extrapolated from place inside me among these musics I the first eight partials of the determin- enjoy and assimilate. In TT:TGE this Third-and finally-I happiest in istic analysis, mapping each frequency became a subtle but integral part of my am my work when I am exploring fresh to its nearest 12-tone-per-octave equal- modeling process, a kind of hybrid syn- concepts and new materials. If my ex- tempered pitch, then transporting an thesis of my assimilation of my subjects’ plorations take me to a rich vein of octave lower to the choral vocal range, musics, their compositional attitudes musical material, I experiment in- shown in Table 1. and approaches, and my own. tensely with the possibilities. The mate- I am particularly attracted to the tim- I think the person who has thefun is rial instructs me in how to form the bral fusion achieved when the choral theperjbrnw. piece, or even a series of pieces with sonorities and the matching sonic fury kindred properties. I learn from all, are heard in rhythmic/harmonic uni- -Max Mathews assimilating the musical implications. I son. Many of the other transformed have also been gratified by the new utterances-especially those where the Second, I am intrigued by the results musical potentialities I have discovered choral setting called for non-pitched, of juxtaposing seemingly disparate for my pieces to take root in and by the speech-like execution-were mapped compositional processes, materials and sometimes surprisingly successful (to ‘by ear’. media, exemplified in much of my work my ear) musical inventions that grow and certainly in TT:TGE: the juxtaposi- from my experiments. Always at the tion of sonorous choral homophony base of this approach is the essence of COMPOSITIONAL and strict canonic procedure in the the piece I am imagining. My explora- APPROACH John Chowning Episode; of nearly un- tions and experiments with technology . . . this dream-vision of an orchestra performable rhythmic complexity with and computer-mediated approaches of loudspeakers . . . those are real a sound-mass thicket of event pattern- come about for musical reasons, not ings in the Jerry Hunt Episode; of par- physical instrumats. . . . Thq be- stemming so much from the newest came my fiends. . . . It’s an act of adox and languor in the Robert Ashley development in music technology for Episode; of quiet spirituality and lively dismery. , . . Rehewhat’s them. . . . its own sake as from musical impera- I don’t compose in a straight line. I interaction in the Pauline Oliveros tives that-it turns out-often cannot always compose around the peri$h- Episode; of the prosaic and enigmatic be achieved without advanced tech- in the David Tudor Episode; of simple 9.. . . I circle it all the time. . , . nology to use as my modus operandi. the boundaries, dealing with the pleasures and inventive genius in the Thus, I am uncomfortable when I am edges. . . transfmrnations . . . Max Mathews Episode; of medium and sometimes introduced as an ‘electronic message in the Morton Subotnick space to sing by itself: composer’. I am simply a composer of Episode. For myself as a composer, music, not an organism interfaced with -David Tudor there is a doubled duality of senti- a machine. On the other hand, that mentality and abstraction contradicted is not necessarily a flawed concept. Three things run through my work as a by irony and corporeality. Dualities I wonder. . . composer and all are present in every fascinate me because they are never detail of 7T:TGE. First-and most im- really understood, just as the polarities portant-I believe that the musics of in the fortunes of life are never fully humankind are concomitantwith aplu- reconciled.

Austin, Booneand Sma, Transmission Two:The Great Excursion 85 frequency trajectories magnitude C >uo magnitude spectra peak detection .-YY .- computation - +- .Y 3 peak continuation magnitude *g g trajectories 32 f -PI window

#waveform

I complex spectra

C Fig. 2. This block diagram of the 0 analysis part of the Spectral Modeling .- Synthesis technique shows how to obtain a deterministic-plus-stochastic envelope envelopes 2 3 representation from an input sound. approximation -5i 2Q WL

TT:TGE AND SPECTRAL terministic’ part) and (2) a time-varying MODELING SYNTHESIS filtered noise component (the ‘stochas- SANSY-THE tic’ part). The SMS technique has SCIENTIST’S When generating musical sounds on a proved to give general, highquality PERSPECTIVE digital computer, it is important to have transformations for a wide variety of a good model with parameters that pro- musical signals. Xauier Serra vide a rich source of meaningful sound transformation. Good examples of pa- The piece T1‘:TCX by Larry Austin was rameters are pitch, amplitude, bright- The Deterministic-Plus- the first composition to make use of the ness, and bow pressure. Three basic Stochastic Model Spectral Modeling Synthesis (SMS) tech- model types are in prevalent use today: The SMS technique assumes that the nique and of the program SANSY [8]. instrument models, spectrum models input sound to be transformed has a de- This computer system, which 1 de- and abstract models. Instrument mod- terministic component and a stochastic veloped at CCRMA, allows for the els attempt to characterize the param- component. A deterministic signal is manipulation of prerecorded sounds eters of a sound at its source, which traditionally defined as anything that is and was used by Austin to transform might be aviolin, clarinet or vocal tract. not noise (i.e. a perfectly predictable a series of recorded voices into his Spectrum models attempt to describe part, predictable from measurements sonic furies. the parameters of a sound at the basilar over any continuous interval). However, SANSY is a software environment membrane of the ear, discarding infor- in our model the class of deterministic written solely for the purpose of de- mation from the sound spectrum that signals considered is restricted to sums veloping the SMS technique. SANSY is the ear seems to discard. Abstract mod- of sinusoidal components. Each sinu- considered more a research tool than a els, such as Frequency Modulation (FM) soid models an actual sinusoidal com- practical program for musical applica- (the technique used in the Yamaha ponent of the original sound and is tions. Using such a technically daunt- DX7 synthesizers), attempt to provide described by an amplitude and a fre- ing program to create music poses a musically useful parameters in a math- quency function. A stochastic, or noise, special challenge to the composer. ematical formula. signal is fully described by its amplitude SMS models sound at the spectrum and its general frequency character- level. It models time-varying spectra as istics. When a signal is assumed stochas- (1) a collection of sinusoids, controlled tic, it is not necessary to preserve either through time by piecewise linear ampli- the instantaneous phase or the exact tude and frequency envelopes (the ‘de- frequency information.

86 Aucfzn, Boonr and Sara, Transmission Two: The Great Excursion Therefore, the input sound s(t) is the sum of a series of sinusoids plus a noise signal,

R * s(t) =~A,(t)cos[B,(t)] e(t) (1) trajectories + transformations r= 1 - magnitude - where A,(t) and Bdt) are, respectively, trajectories transformations deterministic the instantaneous amplitude and the - signal phase of each sinusoid and e(t) is the - synthesized noise component. spectral The model assumes that the sinu- envelopes soids are stable partials of the sound stochastic and that each one can be characterized transformations by its amplitude and frequency. The instantaneous phase is then taken to coordinates be the integral of the instantaneous generator frequency Wt),and therefore satisfies - window

e,(t) = J0 lor(r)m (2) where ~(t)is the frequency in radians, Fig. 3. This block diagram of the synthesis part of the Spectral Modeling Synthesis techni- and ris the sinusoid number. que shows how to obtain a synthesized sound from the representation that resulted from By assuming that eft) is a stochastic the analysis process. signal, it can be described as filtered white noise,

e(t) =J'h(t,z)u(.r)dz (3) 0 where u(t) is white noise and h(t,z) is the impulse response of a slowly time-varying filter. That is, the residual is mod- eled by the convolution of white noise with a frequency-shaping filter. The filtering of a noise signal can be implemented by multiplying the inverse Fourier transform of the filter- frequency response by a random phase term. This method of filtering a noise signal is used to synthesize the stochastic signal.

Description of the SMS Technique Figures 2 and 3 show the block diagrams for the analysis and synthesis parts of the SMS technique. The first step in the analysis is the derivation of a series of magnitude spectra of the input sound by computing the Fast Fourier Trans form (FFT) of every windowed portion of the waveform, i.e. computation of the Short-Time Fourier Transform (STFT). Both the FFT and STFT are mathematical operations that can con- vert a portion of a sound into frequency spectra. From the series of magnitude spectra the analysis detects prominent peaks in each spectrum. These peaks are then organized into frequency tra- Fig. 4. This example of a typical display interface of the SANSY program shows an jectories by means of a peak continua- original cello sound at the top, three analysiis representations in the middle and a syn- tion algorithm. The relevance of this thesized version of the cello sound at the bottom. algorithm is that it extracts the stable sinusoids present in the original sound (the deterministic component).

Ai&n, Awne and ha,Transmission Two: The Great Excursion 87 The stochastic part of the waveform rowed from Speech and Phonetics Inter- Perfect Lives is 'Blue' Gene Tyranny (aka Robert She@. is calculated by first computing the active Research Environment (SPIRE). SPIRE runs on a particular type of LISP 4. The premiere was broadcast live from a public STFT of the deterministic component, performance at Hertz Hall, University of Cal- in the same way that the STFT of the Machine [lo]. It is a program for inter- ifornia, by radio station KF'FA-FM, Berkeley, Cal- original waveform was obtained, and actively manipulating speech signals ifornia, on 26 February 1990. It was performed by the University of California Chamber Chorus, Phil- then subtracting each magnitude spec- and computations on those signals. lip Brett, conductor: ChrisBrown, keyboards; Chris trum from the corresponding spectrum In addition, it can be used as a basis for Chafe, celletto; and William Winant, Boie Radio of the original waveform. The envelope developing other speech-processing sys- Drum and percussion. The second performance was broadcast live from a public performance at the of each 'residual' spectrum is then tems and can be extended and custom- Merrill Ellis Intermedia Theater, University of derived by performing a line-segment ized by the user to perform specific tasks. North Texas (UNT), by radio station WU-FM, approximation. These envelopes repre- The SANSY environment includes Denton, Texas, on 1 April 1990. It was performed by the UNT A Cappella Choir, Henry Gibbons, sent the stochastic signal. a graphic interface that is completely conductor: Carter Enyeart, cello; Steven Harlos, The deterministic signal, i.e. the configurable, from which the user con- keyboards; and Robert Schietroma, KAT controller sinusoidal component, results from the trols analysis, transformation and syn- and percussion. magnitude and frequency trajectories, thesis processes. A typical SANSY- 5. Spectral Modeling Synthesis was designed for use on the Symbolics LISP Computer. or their transformation, by generating system display shows an original sound, a sine wave for each trajectory (i.e. ad- several analysis representations and 6. Controller-instruments used in the first broadcasts included the Yamaha KX88 and the Roland ditive synthesis). a synthesized version of the original A-80 Musical Interface Digital Interface (MIDI) The stochastic signal is the result sound (Fig. 4). The interface allows the digital keyboard controllers, the Boie Radio Drum (developed by Max Mathews at CCRMA) and the of creating a complex spectrum (i.e. user to modify data, including deter- KAT digital percussion keyboard, and the celletto magnitude and phase spectra) for every ministic and stochastic representations, (developed by Chris Chafe at CCRMA). spectral envelope of the residual, or with graphic tools, as well as algo- 7. See Serra's technical explication starting on its modification, and performing an rithmically. The possibilities for trans- page 86. inverse-STFT (using the overlapadd formation are enormou-limited only 8. See X. Serra, A SystemforSoundAnalysis/Transfor- method to form the final output). The by the imagination of the user. Apart mation/Synthesis Based on a Deterministic plus Stochas- magnitude spectrum is the envelope from supporting the SMS technique, tirDwomposition, doctoral thesis, later published as a technical paper by the Center for Computer itself, and the phase spectrum is gen- SANSY includes other standard tools Research in Music and Acoustics, Stanford Univer- erated by a random-number generator. for analyzing, editing and synthesizing sity, 1989; X. Serra and J. 0. Smith,"Spectral Mod- eling Synthesis: A Sound Analysis System Based on This process corresponds to the filter- musical sounds. SANSY is designed for a Deterministic plus Stochastic Decomposition", ing of white noise by a filter with a use as a research environment rather CompuferMusicJournnl14, No. 4, 12-24 (1990). frequency response equal to the spec- than for musical composition. Musi- 9. SANSY is designed to run on the Symbolics tral envelope. cally it is used as a sound-modification LM-2, with the assistance of the FPS AP-12OB array program, although it is difficult to use processor. it to combine or organize sounds 10. D. Shipman,"Development of Speech Re- in time. Therefore, once a sound has search Software on the MIT LISP Machine", The DESCRIPTIONOF THE Journal ofthe Aruustic Soaeq ofAmerica, 103d meet- been created with SANSY, it has to ing, Chicago, Illinois, U.S.A.,26-30 April 1982. SANSY ENVIRONMENT be sent to a soundediting and mixing SANSY has been developed on a LISP- program for integration into a musical Bibliography machine workstation, making use of composition. Austin, L.,"l.ive-Electronic Music on the Third an array processor for the signal- The state of the SANSY system and Coast". Contmpura? Music him6. Part I ( 1990). processing calculations [9]. A LISP ma- of the SMS technique was, at the time Austin, L., and T. Clark, Leanzingto Compose:Modes, that Austin composed 7T:TGE, still Mataialc and Mod& (Dubuque, IA: Wm. C. Brown, chine is a computer built around the 1989). language LISP, a high-level computer under major development. Currently, the SANSY system is being rewritten so Austin. L., and E. DeLisa,"Modeling Processes of language well suited for the specifica- Musical Invention", Proceedings of the 1987 Inter- tion and manipulation of abstract con- that it might be a useful tool for musi- national Computer Musir Conference, (San Francisco, cepts. An array processor is a piece of cians, the major emphasis being com- CA: Computer Music Association, 1987) pp. 200- 207. hardware designed for mathematical position. This new version will run on computations. This combination of a NeXT computers and be available to Clark, T.,"Coasts: On the Creative Edge with Com- poser Larry Austin", Compu(PrMusicJournall3,No. LISP Machine with an array processor anyone interested. 1,21-35 (1989). has the advantage of combining a Clark, T.,"Duality of Prorcss and Drama in Larry highly sophisticated programming en- References and Notes Austin's Sonata Concertante", Perspectives of Nm Mzlsir23,No. I, 112-125 (1984). vironment with a fast computation 1. M. Subotnick,"Silver Apples of the Moon", LP engine. The SANSY software is written recording, Nonesuch Records H 71 174. Schwartz, E., and I.. Starr."David Tudor", NmGone in a dialect of LISP called ZETA- Dirtionup o/Amm'ran Musir 4 (New York: MacMil- 2. N. Strauss,"Klaus Schoening: Radio You Can Ian PressLtd, 1986) pp. 418-419. LISP, except for the low-level signal- See", EurMagazine 15, No. 3, 14-15 (1990). processing routines, which are written Strawn, J., ed., Digital Audio Signal Processing: An 3. R. Ashley, Per&rt Li7m, videotape recording, Anthology (1.0s Altos, CA William Kaufmann, Inc., in a low-level language using tools bor- Lovely Music, New York, 1988. The keyboardist in 1985).

88 Auctin, Roonpnnd Sma, Transmission Two: The Great Excursion