DocumentCode :
2768602
Title :
High-performance hmm adaptation with joint compensation of additive and convolutive distortions via Vector Taylor Series
Author :
Li, Jinyu ; Deng, Li ; Yu, Dong ; Gong, Yifan ; Acero, Alex
Author_Institution :
Microsoft Corp., Redmond
fYear :
2007
fDate :
9-13 Dec. 2007
Firstpage :
65
Lastpage :
70
Abstract :
In this paper, we present our recent development of a model-domain environment-robust adaptation algorithm, which demonstrates high performance in the standard Aurora 2 speech recognition task. The algorithm consists of two main steps. First, the noise and channel parameters are estimated using a nonlinear environment distortion model in the cepstral domain, the speech recognizer\´s "feedback" information, and the vector-Taylor-series (VTS) linearization technique collectively. Second, the estimated noise and channel parameters are used to adapt the static and dynamic portions of the HMM means and variances. This two-step algorithm enables joint compensation of both additive and convolutive distortions (JAC). In the experimental evaluation using the standard Aurora 2 task, the proposed JAC/VTS algorithm achieves 91.11% accuracy using the clean-trained simple HMM backend as the baseline system for the model adaptation. This represents high recognition performance on this task without discriminative training of the HMM system. Detailed analysis on the experimental results shows that adaptation of the dynamic portion of the HMM mean and variance parameters is critical to the success of our algorithm.
Keywords :
distortion; hidden Markov models; linearisation techniques; speech recognition; telecommunication channels; Aurora 2 speech recognition task; additive distortion; channel parameter; convolutive distortion; high-performance HMM adaptation; linearization technique; model-domain environment-robust adaptation algorithm; nonlinear environment distortion model; speech recognizer feedback information; vector Taylor series; Adaptation model; Cepstral analysis; Hidden Markov models; Nonlinear distortion; Parameter estimation; Speech enhancement; Speech recognition; Standards development; Taylor series; Working environment noise; additive and convolutive distortions; joint compensation; robust ASR; vector Taylor series;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Automatic Speech Recognition & Understanding, 2007. ASRU. IEEE Workshop on
Conference_Location :
Kyoto
Print_ISBN :
978-1-4244-1746-9
Electronic_ISBN :
978-1-4244-1746-9
Type :
conf
DOI :
10.1109/ASRU.2007.4430085
Filename :
4430085
Link To Document :
بازگشت