Exploring Billions of Audio Features

By François Pachet and Pierre Roy.

In the Proceedings of the International Workshop on Content-Based Multimedia Indexing (CBMI 2007), pp. 227–235, Bordeaux, France, June 2007.

Read the paper (PDF)
Publisher record

About the paper

What if an audio classifier could use features constructed for its particular task? This paper explores a large space of candidate features made by combining elementary signal-processing operators.

Abstract summary

Pachet and Roy call these combinations analytical features. The paper examines how search and domain knowledge can make it practical to look for useful features in a space that is too large to enumerate.

Related work

See the Feature Generation project and the later journal article Analytical Features.

Back to Publications