The Timbre Space Mapping for VST Synthesizers (TMS4VSTS) is a system which implements mappings between control interfaces and parameters of sound synthesizers through timbre spaces. The system is implemented in Max/MSP, and it can host any VST-Plugin software synthesizer. The generative mapping computation, based on machine learning algorithms, is implemented in MATLAB, which runs in the background and communicates seamlessly with the Max/MSP via OSC.
The control of sound synthesis parameters with the TMS4VSTS provides two advantages:
- Dimensionality reduction of the control space
- High correlation between control and perceptual characteristics of synthesized sound
The proposed control abstraction, illustrated in the scheme below, hides the numerous technical parameters (synthesis algorithm variables) and expose to users only two or three control dimensions, which represents the coordinates of the synthesizer’s timbre space. Browsing the timbre space of the specific VST users can control simultaneously and expressively an arbitrary number of synthesis parameters. This is enabled by a case specific sound-to-parameters model computed by the TMS4VSTS.
Different synthesizer generates different sound timbres. The variation of the synthesis parameters changes the generated sound, as well as its main perceptual characteristics, in different ways. The unique relationship between parameters and sound of each sound synthesizer is usually complex (nonlinearities, correlations, etc.), unknown to end users. To compute the mapping through timbre spaces, the TMS4VSTS perform a prior analysis of the timbre generated by different combination of selected synthesis parameters (automatic or interactive). The result of this analysis is processed by the mapping computation algorithm (dimansionality resuction and machine learning algorithms), which operates unsupervisedly and provides within few seconds a mapping that can be used and tuned in real-time. The TMS4VSTS also features interactive visualizations of the timbre space.
git clone https://github.com/stefanofasciani/TSM4VSTS
svn checkout https://github.com/stefanofasciani/TSM4VSTS
S. Fasciani, “Interactive computation of timbre spaces for sound synthesis control”, in Proceedings of the 2nd International Symposium on Sound and Interactivity (SI15), Singapore, 2015. [PDF]