Speech Recognition Using Features Based on Glottal Sound Source

Abstract

We discussed the adoption of the features based on the glottal sound source for speaker independent speech recognition. It has been thought that such features cannot contribute to the speaker independent speech recognition because of the dominant speaker dependent factor.
In this paper, we tried to use pitch, power, LPC residual power, voicing rate, and their regression coefficients as feature parameters for speaker independent speech recognition and found that some features could improved the performance. We discussed the reason of this contribution and concluded that the procedure to derive the regression parameters could reduce the speaker dependent factor which appeared as biases of those features.
We also tested the parameters on a large-vocabulary continuous speech recognition task and obtained the performance improvement.