Title :
Channel identification and signal spectrum estimation for robust automatic speech recognition
Author_Institution :
Dept. of Comput. Eng. & Comput. Sci., Missouri Univ., Columbia, MO, USA
Abstract :
A feature estimation technique is proposed for speech signals that are corrupted by both additive and convolutive noises via combining channel identification with power spectrum estimation. A correlation-matching algorithm is developed for channel identification, and a Gaussian mixture density model of speech DFT spectra is formulated for estimation of speech power spectra. Cepstral features of speech are calculated from the estimated power spectra. Using the proposed method, significantly improved accuracy was achieved on speaker-independent continuous speech recognition where the speech data were corrupted by a simulated linear distortion channel and additive white noise.
Keywords :
acoustic convolution; cepstral analysis; correlation methods; discrete Fourier transforms; parameter estimation; spectral analysis; speech recognition; white noise; DFT spectra; Gaussian mixture density model; additive noise; additive white noise; cepstral features; channel identification; convolutive noise; correlation-matching algorithm; feature estimation technique; power spectrum estimation; robust automatic speech recognition; signal spectrum estimation; simulated linear distortion channel; speaker-independent continuous speech recognition; speech signals; Acoustic distortion; Additive noise; Automatic speech recognition; Cepstral analysis; Discrete Fourier transforms; Robustness; Signal processing; Spectral analysis; Speech processing; Speech recognition;
Journal_Title :
Signal Processing Letters, IEEE
