Finding Good Acoustic Features for Parrot Vocalizations: The Feature Generation Approach

By Nicolas Giret, Pierre Roy, Aurélie Albert, François Pachet, Michel Kreutzer, and Dalila Bovet.

The Journal of the Acoustical Society of America, 129(2), pp. 1089–1099, February 2011.

Read the paper (PDF)
Publisher record

About the paper

This study applies audio feature generation to the calls of African grey parrots. The goal is to find acoustic measurements that help group recordings into meaningful call types.

Abstract summary

The authors compare the automatic classification with an expert’s judgments based on spectrograms and with other automatic methods. The generated features produced classifications close to the expert’s and improved on the other automated approaches tested.

Related work

See Animal Communication for the wider work on vocalizations, and Analytical Features and the Feature Generation project for the underlying audio methods.

Back to Publications