Title :
Noisy speech recognition via wavelet transform enhancement
Author :
Moon, Seokyong ; Hwang, Jenq-Neng
Author_Institution :
Dept. of Electr. Eng., Washington Univ., Seattle, WA, USA
Abstract :
A speech enhancement system based on discrete wavelet transform (DWT) representation and solved by the estimation maximization algorithm is proposed. This technique can find its counterpart in Wiener filtering where Fourier transform or all-pole modeling representation are used. Very encouraging enhancement results based on the proposed DTW generalized Wiener filtering were obtained, using a time-delay neural network for recognition
Keywords :
filtering and prediction theory; neural nets; speech recognition; wavelet transforms; DTW; Fourier transform; Wiener filtering; all-pole modeling; discrete wavelet transform; estimation maximization algorithm; noisy speech; speech enhancement system; speech recognition; time-delay neural network; Background noise; Delay effects; Discrete wavelet transforms; Neural networks; Noise figure; Speech analysis; Speech enhancement; Speech recognition; Wavelet transforms; Working environment noise;
Conference_Titel :
Signals, Systems and Computers, 1992. 1992 Conference Record of The Twenty-Sixth Asilomar Conference on
Conference_Location :
Pacific Grove, CA
Print_ISBN :
0-8186-3160-0
DOI :
10.1109/ACSSC.1992.269131
