Victor ROSI, PHD Student in the PDS Team at the STMS Laboratory ((Ircam/CNRS/Sorbonne Université/Ministère de la Culture), has presented:
Title: Explaining sound concepts' shared and specific representations
You can listen to his conference on Ircam media channel: https://medias.ircam.fr/x266d2a_victor-rosi
abstract:
The common understanding of concepts used to qualify sounds is crucial for communication between music and sound professionals. Yet, one can wonder how clearly such concepts are acoustically defined and whether sound expertise plays a role in the development of corresponding mental representations. To investigate these questions, we assessed the acoustic fingerprint of four sound concepts (brightness, warmth, roundness, and roughness) in three participant groups (sound engineers, conductors, and non-experts). Hence, participants were asked to rate a database of orchestral instrument sounds (N=520) with a Best-Worst Scaling procedure. This data-driven method allowed us to sort the sound database for each sound concept and professional category. We compared these measures across populations and ran machine learning algorithms to unveil the contribution of acoustic features to each concept and population. Our results reveal the shared and specific aspects of each concept across populations. For instance, the concept of brightness, which is the most used by sound and music experts, is the least shared between populations. Conversely, the concept of roughness, which is not widely used, has a strong shared representation. We present the acoustic similarities and discrepancies of the concepts studied between population, and give an acoustic description of each sound concept for each population.