Supervision: Thach Pham Vu,Vincent Grimaldi

Project type: Semester project (bachelor) Semester project (master)


Granular synthesis is an alternative to conventional sound synthesis (additive, substractive or frequency modulation, among others), not relying on manipulating waveforms, but rather splitting it into sound "building blocks" (grains) instead. This method might be applied to musical composition, but also environmental sound design (especially in gaming, where real-time synthesis is mandatory).

In the frame of a project aiming at automatically identifying annoying car noise in real time, we aim at identifying the characteristics of car noise that are strongly related to annoyance to the public, through a thorough listening-test. In this perspective, a real-time car sound synthesis tool needs to be developed.

The proposed project aims at developing a sound synthesis tool based on granular synthesis, allowing real-time interactions with listeners in the context of listening test. The work will consist in:

  • Setting a strategy for granular sound synthesis of car noises

  • Analyzing a database of recorded car noise, with different characteristics

  • Developping the granular synthesis tool

  • Apply this to a real-time listening test protocol for annoyance assessment

Context: Sound analysis (25%), Sound synthesis (50%), listening tests (25%)