INTERSPEECH 2011 12th Annual Conference of the International Speech Communication Association Florence, Italy August 27-31. 2011

Alternative Frequency Scale Cepstral Coefficient for Robust Sound Event Recognition

Yi Ren Leng (1), Huy Dat Tran (1), Norihide Kitaoka (2), Haizhou Li (1)

(1) A*STAR, Singapore
(2) Nagoya University, Japan

There are two issues when applying MFCC for sound event recognition: 1) sound events have a broader spectral range than speech thus the log-frequency scale is less informative; 2) low frequency noise is more prevalent thus the log-frequency scale captures more noise. To address these issues, we study two alternative frequency scales and show that they outperform MFCCs for sound event recognition under mismatch conditions using Support Vector Machines (SVMs) without the need for complex algorithms.

Full Paper

Bibliographic reference.  Leng, Yi Ren / Tran, Huy Dat / Kitaoka, Norihide / Li, Haizhou (2011): "Alternative frequency scale cepstral coefficient for robust sound event recognition", In INTERSPEECH-2011, 297-300.