LOOPS: An Informal Timbre Experiment




Robert Erickson



Musical structures, no matter what their particular musical dimensions, rest upon perceptions of similarity and difference.   We know something of pitch structure, less about rhythmic structure and almost nothing about timbre as a structural element in music.  Indeed, it has commonly been thought to be a subsidiary and/or ornamental musical characteristic, without potential for generating structure.  Nevertheless, there are compositions by Webern, Varèse and a few other composers which indicate that the prevailing view may be wrong.  LOOPS is an experiment to obtain information about the question: can there be specifically timbral organization in music.


The notion of a structured sequence of sounds whose organization is not primarily a matter of pitch was first proposed by Schoenberg (1911) in the final paragraphs of his Harmonielehre. 


I cannot readily admit that there is such a difference, as is usually expressed, between timbre and pitch. It is my opinion that the sound becomes noticeable through its timbre and one of its dimensions is pitch.  In other words: the larger realm is the timbre, whereas the pitch is one of the smaller provinces.  The pitch is nothing but timbre measured in one direction.  If it is possible to make compositional structures from timbres which differ according to height, [pitch] structures which we call melodies, sequences producing an effect similar to thought, then it must also be possible to create such sequences from the timbres of the other dimension from what we normally and simply call timbre.  Such sequences would work with an inherent logic, equivalent to the kind of logic which is effective in the melodies based on pitch.  All this seems a fantasy of the future, which it probably is.  Yet I am firmly convinced that it can be realized.1


But Schoenberg composed no klangfarbenmelodie himself, and in a letter of 1951 to Rufer (1969) he discussed the idea again, this time with a slightly different emphasis.  After remarking that "Webern's compositions realized only the smallest part of my conception of "klangfarbenmelodie" he goes on to say that what might appear to be klangfarbenmelodie in his own works is always some kind of polyphony, and that these isolated instances are not really melodies.  Melodies require a constructive unity, an organization, and "in my conception such (klangfarbenmelodie) forms had become something new for which as yet there is no description, because, indeed, they do not yet exist."


The remark about polyphony is especially interesting, because it is clear that klangfarbenmelodie must be compound, i.e., exhibit more or less easily perceptible polyphonic characteristics, since the constituent timbres tend to cohere on the basis of timbre class. The reason it is difficult to find unequivocal examples of klangfarbenmelodie  (in Schoenberg's 1911 sense of the term) may therefore be related to the tendency of linear sequences to break into the separate perceptual streams or channels which are implicit in extended melodies as "hidden polyphony" or compound melodic line.


The formation of perceptual streams or channels was first investigated by Miller and Heise (1950).  They found that a rapid trill broke into two separate streams when the frequency distance exceeded about one seventh of an octave in music ranges.  They named the region of transition the "trill threshold". Norman (1966, 1967)  commented upon the experiment of Miller and Heise, offering several possible explanations of the phenomenon, and reported that when listeners were asked to decided whether a probe tone followed the higher or lower of two repeating background tones, they had little difficulty when the frequency of the probe tone was between those of the two background tones, but when the probe tone was much higher or lower than the background tones, the task was correspondingly more difficulty.  Warren and his associates (1969) found that in listening to a tape loop of four sounds (40 Hz. square wave, 1000Hz sine wave, the vowel "ee", white noise burst) played at a rate of 200 msec for each of the four items, their subjects had great difficulty in judging the order of the sounds.  The four different sounds tended to form four separate perceptual channels unless played very slowly.  Bregman and Campbell (1971) suggested that stream formation is a primary auditory phenomenon, and Bregman and his associates have performed a number of later experiments related to the perception of melodic patterns in musical situations.


My LOOPS experiment is an attempt to discover whether concepts of perceptual stream or channel can be helpful in a musical understanding of the effects of fast discrete changes of instrumental timbre.  If a repeating melodic pattern of six pitches is performed by five instruments, with each instrument playing a single note, in the manner of a hocket, then each pitch of the pattern will eventually be played by a different instrument (Example 1) and one can form opinions about the effects of the timbral versus the pitch dimension.






Clearly one is able to listen to this delicious confusion in more than one way: (1) one may follow the tune through its changes of timbre; (2) one may begin to form perceptual streams on a pitch basis (in this kind of listening the C/B flat patterns of the high line and the E flat/D flat patterns of the low line are clearest); (3) one may follow the line of a single instrument (marimba is easy, clarinet is more difficult); (4) one may listen - and this is most likely  - in a mixed manner, using (1) or (2) or (3), depending upon the detailed musical situation at any particular moment.


With fewer instruments one might expect the channeling to be stronger, because each instrument is heard with fewer rests between appearances.  Example 2 employs three instruments at I and five instruments at J in an eight pitch melodic pattern.




It may be very slightly easier to follow an individual instrument line when three instruments are employed, but not much.  Following an instrument appears to depend more upon the special characteristics of the instrument, especially its attack quality, in relation to the total group or subgroup.  The marimba line can be tracked almost too easily, but the other two instruments at I are not so easy to disentangle.


If a melodic pattern has a wider pitch ambit, Example 3, one could expect more channeling on a pitch basis.  Example 3 also uses a more complex pattern:  a nine tone sequence plus a nine tone variant (simple repeating sequences soon become boring) with enough octave displacements to make a three octave range.  A further complication here is that instead of simply rotating the instruments in sequence they are used in patterns.  Sometimes the pattern is modified because a pitch is not available on a particular instrument. Especially interesting is a noticeable change of grouping which must be timbre determined -at O, after the complex pattern of N and before the complex pattern of P.














If one lengthens notes in one of the instruments a melodic channel in the timbre of the instrument will be formed.  There is nothing musically new or problematical  in passages such as are shown in Example 4, although perceptual channeling becomes very complex, and one is very aware of texture and of the intensified disjunction (contrast) between the instrumental timbres.  Notice too, that the pitch range has been further extended.









        It appears that the answer to the question, can a melodic figure be preserved while undergoing radical changes of instrumental timbre is yes - but.  The "but" includes matters such as the total range of the melodic pattern, the tempo, the particular instruments involved, their timbres at the specified pitches and the type of articulation of the attack and decay.  It is easy, but allowing only slight overlaps of sound, to turn a precariously sequential melodic formation into something clearly polyphonic.


How strong is the effect of the timbre pattern vis a vis the melodic pattern?  Strong, but no general statement is possible.  The sub-patterns produced by the competition between pitch channeling and timbre channeling are local effects, but controllable, and full of compositional potential, not least in the area of rhythm.  It is of great interest that, in spite of the meter of the counting process, the patterning is chiefly a result of tonic accent, loudness of the various notes in an instrument's repertory, articulation of attack and decay, etc.  Timbral distinctiveness or vividness in any micro context sounds very important, perhaps crucial, to the formation of the sub-groupings, and, therefore, the rhythm and the higher levels of the musical organization.


Why is the number of effective perceptual categories often less than six, even when all instruments are playing?  The answer to this is an interesting problem in psychoacoustics (are psychologists listening?) but it is of musical interest too, for the confusions among clarinet, saxophone and bassoon in certain contexts of LOOPS mean that there are different perceptual contrast relationships in different musical situations.  The distinctiveness of a timbre, and therefore its contrast potential, is different in different registers (or even different pitches) in a non-simple way.  we cannot think merely in terms of gross contrast - clarinet, trumpet, saxophone, etc., but most always consider the timbre of the instrument at whatever particular pitch it is playing.  Now, if certain instruments can be composed in such a way that either they can be made to tend toward homogeneity or confusability of sound or toward diversity and distinctiveness of sound, then there is a possibility for a structural interplay between timbre and pitch, and that is the most important musical insight to be gained from LOOPS.




I wish to thank the LOOPS players, who rehearsed a difficult musical score for many hours over a period of months.  They included Mel Warner, clarinet, Edwin Harkins, trumpet, Jean-Charles Francois, marimba and Ron Grun, bassoon, from the Project for Music Experiment, UCSD, and Larry Livingston alto saxophone and Peter Middleton, flute, from the Department of Music, UCSD.  Charles White, from the Project for Music Experiment, was the recordist.



Encinitas, March 26, 1973.






Bregman, A.S. and Campbell, J. (1971): "Primary Auditory Stream Segregation and Perception of Order in Rapid Sequences of Tones," Journal of Experimental Psychology Vol. 89, no.2, 244-249.


Miller, B.A. and Heise, G.A. (1950): "The Trill Threshold," Journal of the Acoustical Society of America Vol. 22, 637-638.


Norman, D.A. (1966): "Rhythmic Fission: Observations On Attention, Temporal Judgements and the Critical Band," unpublished ms. Harvard University, 1966.


Norman, D.A. (1967): "Temporal Confusions and Limited Capactiy Processors, Acta Psychologica, Vol.27, 293-297.


Rufer, J. (1969):  "Noch Einmal Schoenbergs Op.16," Melos, 367.


Schoenberg, A. (1911):  Harmonielehre, U.E. No.3370 Leipzig/Wien, p.471.


Warren, R.M.; Obusek, C.; Farmer, R.; Warren, R.P. (1969): "Auditory Sequence: Confusion of Patterns other than Speech or Music," Science Vol. 164, 2 May, 536-537, .



1 Schoenberg, Harmonielehre Leipzig and Vienna, 191, pp 470-471.