Speech as Timbre Models for Orchestration – A Comparative Study Between Cantonese and Québécois French

Abstract:

The proposed research-creation project aimed at exploring an alternative conception of orchestration based on natural language timbral imitation. This idea arose from the participants seed idea presented during the CORE Ensemble Project 2021-2022. While one was interested in timbral imitation and emergent timbre, the other used the inflections of Cantonese language to derive musical material. The proposed project aimed at combining these two ideas into one, using a more science-supported approach this time.

The project was rooted in target-based instrumental synthesis, where combinations of instruments and techniques is determined in order to mimic a sound target as closely as possible. While this type of technique was pioneered long ago by composers such as Tristan Murail, the specificity of the proposed project was that the targets were not natural sound phenomena or instruments, but specifically speech.

It has been shown that speech synthesis in a strict sense is very difficult to achieve, even using computer synthesis, therefore the goal was mainly to have this idea stimulate instrumental timbral research and provide a framework within which to think differently about orchestration. A comparative approach was used as one composer will work on Cantonese language, and the other on Québécois French.

The goal was first to conduct research about instrumental combinations based on phonemes as targets. Then each composer wrote a short piece for small ensemble based on the idea of timbral imitation of natural language.