Abstract

Most of the research on Computational Audio Analysis has been on classifying the surface phenomena associated with acoustic signals and with speech events. The meaning of these events usually depends on the context in which they occur. The analysis of audio (and video) scenes can help machines to interpret speech of humans or of human-machine interactions. One of the important issues is how to decide which contextual information to acquire and how to incorporate it into machine learning. Machines should be able to deal with interactions with multi-speakers and interpret the relationship between speakers. To give to the machines the capabilities to interpret and generate appropriate signals taking into account the context of the interaction (with multi-sources analysis) is a real challenge.

Date

2014

Authors

Meinard Müller, Shrikanth S Narayanan, Björn Schuller

Journal

Computational Audio Analysis

Pages

17