You are here


Content-based Audio Transformation
Dec 2000 - Dec 2003
Project description

This project will study and develop a support tool for musical recording manipulation with the goal of enhancing sound and musical qualities in professional sound post production. Specifically, the project will address the description of musical attributes, music database storage and retrieval of those descriptions, and sound transformation based on descriptions using physical attributes (e.g. pitch), perceptual attributes (e.g. brightness), and musical knowledge (e.g. tension).

The components used for sound content analysis and description, as well as the components for physical manipulation of sound, are based on spectral modelling analysis and synthesis techniques. Moreover, the descriptions to be used will be compatible as far as possible with the future MPEG-7 standard for multimedia content description. The transformation components will use both case-based reasoning techniques and modelling of musical knowledge.

Research Line
Machine Learning for Music