Timbre Spaces: New Graphical Models for Analysis and Composition

Recent work in diverse fields of music scholarship, perception, and composition has shown an intensified interest in timbral organization and hearing. But only recently have technological tools made possible an interactive platform for exploring such “timbre spaces” analytically and compositionally. In this paper I will show that recent advances in digital audio feature mapping can be used advantageously to model timbre space for both analytical and compositional purposes. At the heart of this method is corpus-based concatenative sound synthesis (CBCT) as implemented with the computer program CATART, a package of tools developed by Diemo Schwarz for the computer program Max/MSP with the FTM&Co. signal processing library. Such timbre space visualizations can be used to identify analytical results difficult from the notated score alone, or even from a spectrogram, as demonstrated in examples from Debussy, Ravel, and Grisey.
A similar “timbre map” may be used by a composer to plan a form in advance, as harmonic or rhythmic graphs have served composers in the past. Examples include using a timbre space representation to organize the sequence of recorded materials in a work. Smaller-scale musical materials can also be organized using a technique of corpus-based transcription (CBT), where a target trajectory in timbre-space and time, generated from an analysis of an audio recording, is mapped to a recorded corpus. If the corpus comprises acoustic instrumental samples, these may then be transcribed to a score using the computer program OpenMusic.