DocumentCode :
339161
Title :
New features based on the Cohen´s class of bilinear time-frequency representations for speech recognition
Author :
Chen, Jingdong ; Xu, Bo ; Huang, Taiyi
Author_Institution :
Nat. Lab. of Pattern Recognition, Acad. Sinica, Beijing, China
fYear :
1998
fDate :
1998
Firstpage :
674
Abstract :
Although short-time Fourier analysis-based features such as LPCC and MFCC have been widely used in state-of-the-art speech recognizers, the short-time analysis technique suffers from the well-known trade-off between time and frequency resolution and works under the assumption that a speech signal is short-time stationary. This paper investigates an approach using Cohen´s class of bilinear time-frequency distributions representing a speech signal for speech recognition. Preliminary experiments show that the new feature can better represent speech signals and can improve the accuracy of a speech recognizer
Keywords :
cepstral analysis; signal representation; speech recognition; time-frequency analysis; Cohen class; accuracy; bilinear time-frequency representations; cepstrum; speech recognition; Kernel; Laboratories; Mel frequency cepstral coefficient; Signal analysis; Signal processing; Signal resolution; Speech analysis; Speech processing; Speech recognition; Time frequency analysis;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Signal Processing Proceedings, 1998. ICSP '98. 1998 Fourth International Conference on
Conference_Location :
Beijing
Print_ISBN :
0-7803-4325-5
Type :
conf
DOI :
10.1109/ICOSP.1998.770301
Filename :
770301
Link To Document :
بازگشت