Robust Speech Recognition in a Car Using a Microphone Array

Performance of automatic speech recognition relies on a vast amount of training speech data mostly recorded with little or no background noise. The performance degrades significantly with existence of background noise, which increases type mismatch between train and test environments. Speech enhancement techniques can reduce the amount of type mismatch by extracting reliable speech features from the background noise. At very low SNR with nonstationary noise, the enhanced speech may still contain significant noise either in noise-only segments or speech segments. The former masquerade as nonexistent speech features and the latter as distorted speech features. Both significantly degrade the performance of the automatic speech recognizer. This encourages the use of voice activity detection (VAD) algorithms to determine parts with speech present. To use only the reliable speech features, we need to further determine whether the features from the speech region are mainly from speech or from nonstationary noises masking the speech. For more robust speech recognition, this thesis proposes a three-hypothesis VAD consisting of H0: noise-only region; HS : speech-dominant speech region; and HN : noise-dominant speech region. Spectrum-based VAD uses knowledge of the noise spectrum to detect voice activity using the nonstationary nature of speech. This thesis proposes a method of estimating the instantaneous noise spectrum for VAD at low SNR when the noise estimation becomes inaccurate due to its high variance. The spectrum-based VAD, however, cannot distinguish speech from nonstationary noise because both appear nonstationary to the VAD, and thus look like speech. A microphone array can determine the noise-corrupted speech region when the nonstationary