A new speech feature insensitive to the variation of different speakers

A novel robust speech feature which is based on the modified Mellin transform is proposed in this paper. Because of the scale invariance property of the modified Mellin transform, the new feature is insensitive to the variation of the vocal tract length among individual speakers, and thus it is more appropriate for speaker-independent speech recognition than the popularly used melscale frequency cepstral coefficients (MFCC). Experiment has been performed and the result shows that, in comparison with the MFCC, the new feature is able to not only improve the performance of a speaker-independent speech recognizer effectively, but also greatly reduce the standard deviation of the error rates for different outlier speakers.