Real-Time Uses of Low Level Sound Descriptors as Event Detection Functions

Abstract

This paper is a continuation of the research and development we began last year on the study and the use of audio descriptors for real-time performance of mixed acoustic and electroacoustic music, and as tools for computer-assisted musical analysis. In this paper, our main goal is to propose easy and efficient strategies and methodologies for event detection in this context. We will examine three cases of the use of audio descriptors to build event detection functions: spectral slope, spectral standard deviation and the construction of compound descriptors.

Acknowledgements

We would like to thank Gregory Taylor and David Coll for their valuable remarks, comments and suggestions.

Notes

¹All the techniques presented in this paper where implemented using the Zsa.descriptors library (http://www.e-j.com) in the Max/MSP environment (© Cycling ’74, http://www.cycling74.com/).