Title :
Residual noise compensation by a sequential EM algorithm
Author :
Yao, Kaisheng ; Shi, Bertram E. ; Cao, Zhigang ; Wang, Yongqian
Author_Institution :
Dept. of Electron. Eng., Tsinghua Univ., Beijing, China
Abstract :
We present a sequential EM algorithm for residual noise compensation. Residual noise is estimated sequentially by the algorithm, and the estimated noise is used to update the linearized likelihood for noise compensation. Experimental results show that the algorithm can improve the robustness of the recognition system to additive noise
Keywords :
noise; optimisation; sequential estimation; speech processing; speech recognition; statistical analysis; additive noise; expectation maximization algorithm; global noise mean; linearized likelihood; noise statistics; noisy speech processing; residual noise compensation; residual noise estimation; sequential EM algorithm; speech recognition system; Additive noise; Digital communication; Equations; Filter bank; Laboratories; Microwave technology; Noise robustness; Speech enhancement; Speech processing; Statistics;
Conference_Titel :
Signal Processing Proceedings, 2000. WCCC-ICSP 2000. 5th International Conference on
Conference_Location :
Beijing
Print_ISBN :
0-7803-5747-7
DOI :
10.1109/ICOSP.2000.891622
