DocumentCode :
3084509
Title :
Speaker Independent Recognition on OLLO French Corpus by Using Different Features
Author :
Huang, Lixia ; Zhang, Xueying ; Evangelista, Gianpaolo
Author_Institution :
Inf. & Eng. Dept., Taiyuan Univ. of Technol., Taiyuan, China
fYear :
2010
fDate :
17-19 Sept. 2010
Firstpage :
332
Lastpage :
335
Abstract :
The Oldenburg LOgatome speech corpus (OLLO) is specifically designed for evaluating speech recognition methods on variability. The performance of features carried on intrinsic variabilities in speech is meaningful for automatic speech recognition (ASR) system. ZCPA and MFCC were the two main features applied to OLLO French corpus in this paper. We took cepstral mean subtraction (CMS) on MFCC. Dynamic transforms (delta-delta-ZCPA and delta-delta-MFCC) were also adopted. The experiments show that the MFCC outperform the ZCPA in separate style. But ZCPA is more robust between different variabilities. The delta-delta operation of MFCC achieves best recognition in noise-free environment. Moreover, ZCPA could be complementary to MFCC so that one can combine them together especially on soft speaking style.
Keywords :
cepstral analysis; feature extraction; speaker recognition; transforms; MFCC; OLLO French corpus; ZCPA; cepstral mean subtraction; dynamic transform; oldenburg logatome speech corpus; speaker independent recognition; speech recognition methods evaluation; Feature extraction; Filter bank; Hidden Markov models; Mel frequency cepstral coefficient; Robustness; Speech; Speech recognition; MFCC; OLLO corpus; ZCPA; delta-delta-MFCC;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Pervasive Computing Signal Processing and Applications (PCSPA), 2010 First International Conference on
Conference_Location :
Harbin
Print_ISBN :
978-1-4244-8043-2
Electronic_ISBN :
978-0-7695-4180-8
Type :
conf
DOI :
10.1109/PCSPA.2010.87
Filename :
5635706
Link To Document :
بازگشت