Title :
Binaural Bark subband preprocessing of nonstationary signals for noise robust speech feature extraction
Author_Institution :
BMW AG, Res. & Dev. Centre, Munich, Germany
Abstract :
A two channel approach to noise robust feature extraction for speech recognition in the car is proposed. The coherence function within the Bark subbands of the mel-frequency-cepstral-transform is calculated to estimate the spectral similarity of two statistic processes. It is illustrated how the coherence of speech in binaural signals is used to increase the robustness against incoherent noise. The introduced preprocessing method of nonstationary signals in two microphones results in an additive correction term of the mel-frequency-cepstral-coefficients
Keywords :
acoustic noise; cepstral analysis; feature extraction; hidden Markov models; microphones; speech recognition; statistical analysis; transforms; HMM; additive correction term; binaural Bark subband preprocessing; car; coherence function; incoherent noise; mel-frequency-cepstral-coefficients; mel-frequency-cepstral-transform; microphones; noise robust speech feature extraction; nonstationary signals; spectral similarity estimation; speech coherence; speech recognition; statistic processes; two channel approach; Acoustic noise; Feature extraction; Frequency estimation; Hidden Markov models; Microphones; Noise reduction; Noise robustness; Speech enhancement; Speech recognition; Working environment noise;
Conference_Titel :
Time-Frequency and Time-Scale Analysis, 1998. Proceedings of the IEEE-SP International Symposium on
Conference_Location :
Pittsburgh, PA
Print_ISBN :
0-7803-5073-1
DOI :
10.1109/TFSA.1998.721498
