For the 2024 Timbre and Orchestration Summer School at the University of British Columbia.

  • The Timbrenauts project report on the Timbre and Orchestration Resource website: [link]
  • The sounds used in the perceptual experiment appear in the above TOR article as well: [link]
  • The interactive Multidimensional Similarity Space (MSS – figure Axyz on the poster): [link]
  • The interactive Closest Timbral Neighbors graph (CTN – figure Bxy on the poster): [link]
  • The interactive Greatest Timbral Diversity graph (GTD – figure Cxy on the poster): [link]
  • The poster as presented at the conference: [link]


Providing Timbre Maps for Musicians Based on Similarity Judgements of an Instrumental Duo’s Extended Techniques

Contemporary musical practice involves extended modes of instrumental sound production, which allow performers to produce a wide range of different timbres. Timbral dis/similarity is a key consideration in orchestration, defining the perception of the resultant auditory scene by affecting concurrent and sequential grouping, as well as formal segmentation. In orchestration research, timbre spaces based on perceptual similarity judgement data have historically focused on comparing standard playing techniques between large groups of instruments—i.e. the instruments of the western classical orchestra playing a short pitched sound without the use of extended techniques—collapsing each instrument to a single point in the timbre space. Following a recent experiment providing perceptual similarity data on a set of extended techniques recorded by a trombone and cello instrumental duo, we can provide potentially more robust information to musicians wanting to create music with this instrumental duo using these instruments’ extended sound palettes to achieve diverse orchestrational goals. This poster presentation focuses on various different ways to present this data visually. These include a similarity space created using multi-dimensional scaling, audio feature spaces created from interpreting the previous, and network graphs connecting recorded samples based on whether they score above or below certain similarity thresholds. The presentation has two goals: to elicit feedback on clarity and usability of these maps, and to entice potential collaborators for future creative output based on the data being presented. This is a continuation of the Timbrenauts student collaboration project awarded previously by ACTOR.