Summary
International Technical Conference on Circuits/Systems, Computers and Communications
2008
Session Number:P1
Session:
Number:P1-102
Robust Speech Recognition Features Based on Temporal Trajectory Filtering and Non-Uniform Spectral Compression
Sang-Ho Lee, Jeong-Hyun Ha, Woo-Young Lee, Jae-Keun Hong,
pp.-
Publication Date:2008/7/7
Online ISSN:2188-5079
DOI:10.34385/proc.39.P1-102
PDF download (97.2KB)
Summary:
This paper proposes a new feature extraction method based on temporal trajectory filtering and non-uniform spectral compression and examines its performance with two tasks in noisy environments. Temporal trajectory filtering is effective for robust speech recognition in noisy environments, due to human hearing is more sensitive to relative values rather than absolute values and the effect of additive noise which varies slowly may be removed. However, even if noise is stationary, it is not removed exactly due to the random fluctuation. Thus we use non-uniform spectral compression after temporal trajectory filtering and this method shows better performances than the respective methods.