Car dif f Univers ity

School of Phys ics

Lecture: Sound Synthesis
Cours e Conv enor: Profess or Mike Greenhough
Spring Semes ter 2005

Principles and techniques of Spectral music

Written by:
Justin Lepany
Senghennydd Cour t
D-6- 1
Salisbury Road
CF 24 4 DS

Sc hool of Mus ic
Student Number :
Content

1. Introduct ion

2. Timb re, spectru m and envelope

3. Instru mental synthesis and its different models

4. Formalization o f spect ral music

5. Conclus ion

A ppendix

Referen ces

2
1. Introduction

Spectral m usic – also called spectralism - appeared in the ear ly 1970’s. To descr ibe it in a

short an d ro ugh way, it is a m usic based on the soun d itself. It was created as a po ssible solution to

problem s of serial music and electronic music. For many composers, ser ial m usic of that tim e was

too strict and abstract: the concept ualization was steady but the result was com pletely unpredictable

and the listener’s per ception was th us neglected, as well as sonority, timbre. On the opposite,

electronic music of that time was of cour se based directly on soun ds but lacked som e writin g and

organization principles due to the use of “already made” so unds. The spectralists’ goal was to

create a music that wo uld reconcile sonor ity and formalization. By the way, spectral music was to

rehabilitate conson ance without looking back to the tonal era. This issue was ho wever discussed

from the 1960’s on and some composers as fam ous as Gyor gy Ligeti, Iannis Xen akis or Krzysztof

Pender ecki had alr eady been trying to fin d a balance bet ween so un d an d conceptualization.

What started sp ectral m usic off were the gr eat ach ievements of m odern aco ustic sin ce they gave

access to the in side of so unds thro ugh devices as the spectrograph wh ich analyses a soun d an d can

giv e a three-dimen sional image of it, i.e. the frequencies, their dyn amic an d their behavio ur in tim e.

The Sp ectral m usic con cept appeared un der three differ ent form s in differ ent European p laces more

or less at the sam e time, around 1973. The first and m ost fam ous case is the French school

com posed of a few Olivier Messiaen’s p upils whose m ost well-known composers are Gerard

Grisey (1946-1998) an d Tristan Murail ( b. 1947). At the sam e time, a sp ectral tren d appeared in

Germany with the group Feedback (Peter Eötvös, Mesias Maiguashca, etc.) whose members are

mostly som e Karlheinz Stockh ausen’ s discip les. The last trend started off in Rom ania an d its m ain

representative is Horatiu Radulescu ( b. 1942).

3
The present work aim s to present a few essential aspects of composition in spectral m usic, namely

instrum ental synthesis of various so und models an d the use of process musical form. The fir st

section briefly defines tim br e to provide the reader with a few notions necessary to understand

spectral m usic. The secon d section explains in strumental synthesis and the thir d section finally

sho ws what kin d of procedur es spectral m usic composer s use to formalize their pieces.

2. Timbre, s pectrum and envel ope

In Spectral music, the fir st step of the composition process is the analy sis of the so un d with

a spectrograph. Look ing into the spectrogr am, the composer can seize the characteristics of the

tim bre.

Tim bre (Lat.: tonu s) is the tonal quality of a soun d, i.e. what en ables one to distin guish a violin and

a flute playin g the same pitch at the same loudness. A soun d hear d as a pitch is actually an

ensem ble of reson ating frequencies: the fundam ental and the overtones. If the overtones are who le

multiples of the fundamental then they are called harmonics an d the spectr um is harmonic. This is

the case of pitched instrum ents like the horn or the cello. If the overtones are not integer m ultiple of

the fun dam ental then the overtones are called partials an d the sp ectrum is def ined as inharmonic. In

diff erent extend, this is the case of bells, percussion s or noises, from the wav es to artificial white

noise which is a so und con sisting of all audible fr equencies hear d with equal inten sity. The

beh avio ur in time of the spectrum defines the tim bre of a so und.

Another defin ition of timbre is the envelope, which is the shape of a so un d’s amplitude in tim e. An

envelope is characterized by its attack, its steady state an d its decay. For in stance, the env elope of a

violin’ s note f eatur es a lo ud attack due to the bo w to uch in g the string and a very short decay sin ce

the string almost com pletely stops to vibrate as soon as the bo w is not in contact with the string

anymore.

4
These details are all important for the com poser of spectral m usic since his m usic aim s to be as

close as possible to a soun d m odel.

3. Instrumental synthesis and i ts different models

Instrum ental synthesis consists of using the orchestra as a m ean to synthesise soun ds. It

basically features three types of m odels: instr um ental harmonic spectr um, inharmonic spectrum and

electronic music devices. The in strum ent m odel is the most current case: hav ing fir st analysed the

spectr um of a chosen instr ument, the com poser orchestrates the spectr um , i.e. he assigns the

spectr um’s harmonics to every instr uments of the orchestra. This procedure is outstan din g since the

instrum ents are not consider ed as a part of a group anymore (brass, strings, etc.) and are not freely

selected accor ding to the com poser’ s taste or for their cultural connotation (the sheph erd’s flute, the

melancholic o boe, etc.): every single instr ument is chosen accor din g to its ability to sound an

overtone'
s pitch in the best way, i.e. when its ran ge an d its so un d in a specific part of this very

range enables the note to soun d as close as an overtone as po ssible.

After this selection phase, the composer must write the part of every voice in a way that sim ulates

the beh avio ur in time of every harm onics: the dynamic signs, the rhythm s, the durations an d the

rests are selected in that purpo se. In order to be as close to reality as po ssible, this step r equires the

use of very precise notation such as quartertones and eigthtones. An im portant point to be

mentioned is that spectral m usic aims to make an orchestra soun d like a tim bre but not like an exact

phenom enon. A two-secon ds soun d can be the model for a ten minutes piece and therefore will be

streched an d expan ded by the com poser.

The most fam ous example of instrum ental synthesis of an instrum ent so und is Partiels (1975) by

Gérard Grisey. This piece belongs to a six-piece cycle (Pro logue, Périod es, Partiels, Modulation s,

5
Tran sitoires an d Epilogue) written bet ween 1974 an d 1985, which is some kin d of antholo gy of the

spectral music techniques. The beginnin g of this 18-minutes-long piece is based on the simulation

of a trombone low E. On the exam ple 1, one can see that the three first bar s ar e just m ade of a long

low E p layed by the trombone doubled by the do uble bass. Then, after the introduction of the

model itself, starts the orchestral expanded sim ulation of this v ery m odel.

Example 1: the beginn ing of Partiels

Source: Grisey ( 1975) "Partiels", Paris: Ed. Ricor di.

The secon d kind of model is inharmonic spectrum. The steps to follow are the sam e as with

harmonic sp ectrum . An exam ple of m usical synthesis of an inharm onic model is Gond wana (1980)

by Tristan Mur ail in which a bell sp ectrum is used. This p iece is explained with more details in a

further paragr aph.

6
The third m odel for instrumental synthesis is the simulation of electronic dev ices. A com mon case

is the use of amplitude modulation, frequency m odulation and rin g m odulation's algorithm to

gen erate artificial spectrum s. Many other electronic music devices can be taken as mo dels.

Tristan Murail's Mémoires-Erosions (1975-76) for horn and nine instr uments is an outstan din g case

which aims to transcribe feedback. M urail is the only composer who is named in this section of the

present work since he is also well known to be deeply involv ed in live electronic m usic an d, as a

consequence, has com posed som e of the best spectral pieces among those f eatur in g in strumental

synthesis of electronic devices. Back in the early day s of electronic music, f eedback was made with

one tape player an d one tape recorder linked together. A signal wo uld be sent to the tape recorder

which would recor d it an d sen d it to the tape player. The former wo uld then send it back to the tape

recorder which wo uld be feeded in the meanwhile by som e others signals. The result is a soun ds

mixture that gets richer an d richer.

This process is explo ited in Mém oire ('

Mem ory'
), which is the fir st part of the piece. This kind of

feedback was a complex pro cedure: after a while, the tape wo uld start to wear o ut due to an

excessive amo unt of re-recor din gs an d what wo uld com e o ut the tape player wo uld pro gressiv ely

soun d like noise rather than a rich polyphony. The second part of the piece, called Erosion,

sim ulates the described process. App endix 2 sho ws the f irst eight loops of the piece where one can

see the horn playin g diff erent short motifs one after the other. Every motif is then repeated by the

diff erent instr uments as the horn keep s on playin g some new m otifs wh ich ar e also repeated, and so

on.

Apart of this o utstan ding case, many other electronic dev ices h ave been sim ulated : Mémoire-

Ero sion simulates delay as well. Appen dix 3 sho ws that som e rhythm marks are placed belo w

7
every bar for an instrument to simulate a delay effect. In som e other pieces, one can find some

reverb or even som e phaser simulation.

4. Formalizati on of s pectral music

The last great feature of spectral music is its particular vision of the m usical form alization.

Instead of usin g the usual narrative form s like sonata or variations, it explo its the soun d completely

and generates the form from the soun d’ s envelope. Lookin g into the sonogr am of a five-secon ds-

long piano note, one can see that the behavio ur in time of every com ponent is different: for

example, some overtones are fir st dominant, then disappear and app ear again but less loud. The

com poser can expand the gener al shape in duration to get obtain the general form of the piece as

well the climaxes an d the quite moments. The compression in time of the beh aviour of sin gle

com ponents can inspir e the rhythms of an instrum ental line, et cetera. This procedure was used

during the spectralism ’s very fir st year s an d was then often replaced by another on e that had

already been used by com poser lik e Stockhause: processes.

The use of processes in m usic is firstly the definition of t wo states: the prim al one and the final one.

The composition procedure then takes the composer to transform the prim al state into the final one,

making the changes as subtle an d as in discern ible as possible. Tristan Mur ail gives us a clear

example in his piece Gond wana for an ensem ble of wood instrum ents, brass in strum ents and

resonating instrum ents (p iano, vibr aphone and crotales), com posed and perform ed in 1980 (see

appendix 2). In the first part, the prim al m odel is a bell soun d an d the final model is a trumpet note.

In bet ween are ten different interm ediate states of wh ich the transitions from one to the other

slo wly melt the virtual bell sound into a virtual trum pet sound. This transformation has other

underlying meanin gs as this p iece is also the transit from an inharmonic spectrum to a harmonic

spectr um.

8
5. Concl usion

Even though spectralists try to be r igoro us and more or less scientific, there are two f acts

that make any hope of real in strumental synthesis utopian: First, acoustic in struments don'
t produce

sine waves but complex soun ds and therefore can'

t so und overtones. Second, the use of tem pered

instrum ents, rythm ns patterns and dynam ic marks can not ren der a realistic imitation of the pitch,

the loudness and the duration of an overtone, even if the com poser dem ands quartertones or

eighthtones and use some very thorough notations. Anyone who aim s to synthesise timbre with

aco ustic instruments an d h um an m usicians is unfort unatly doom ed to failure.

The french composer Philippe Manoury once p ut the m atter in a nutshell by writing abo ut spectral

music, that “there are not more spectrum in this m usic than in any other m usic” (Deliège 1998,

p.803).

Nevertheless, even if the quest for a n ew m usic entirely in spir ed by the so und itself was a

disapo intem ent, spectral music is still a v ery sp ecial m usic with a very unique aesthetic. Its

mysterious an d slo w sonorities still fascin ate the audiences an d the composer s of today. More than

thirty years after the fir st spectral pieces, sp ectral m usic influen ces many com posers from differ ent

schools. Such a fact is rare eno ugh in the field of contem porary music and to be m entionned as well

as remembered.

9
Appendi x

Appen dix 1 : Cohen-Levinas, Danielle (1998) "L'

itinéraire", Par is: L'Harm attan, p.108.

10
Appen dix 2 : Cohen-Levinas, Danielle (1998) "L'
itinéraire", Par is: L'Harm attan, p.109.

11
References

Barr ière, Jean- Baptiste (1991) "Le timbre, métaphore pour la composition", Pa ris: I.R. C. A.M et

Christian Bour geois Editeur.

Cohen- Levinas, Danielle (1998) "L'itinérair e", Paris: L'

Harmattan.

Deliège, Célestin (2003) " Cin quante ans de m odernité m usicale", Paris: Pierre Mardaga.

Dufourt, Hugues (1991) "Musique, Pouvo ir, Écritur e", Paris: Bo ur geois.

Fich et, Laur ent (1996) Les théories scientifiques de la musique au XI Xe et XXe siècles", Paris: J.

Vrin.

12