| In real world speech processing systems, speech signals are often corrupted by background acoustic noise or reverberation. Additionally, for systems which involve transmission of speech data over error-prone communication channels, signals may suffer from packet loss. This dissertation addresses two general frameworks for which compensation of corruptive acoustic noise and channel errors can benefit performance, namely remote speech communication and automatic speech recognition.;In the case of ASR, front-end missing feature (MF) spectral reconstruction is explored. Two solutions are offered, the first of which uses HMM-based processing and accounts for temporal and/or frequency correlation. The second exploits the sparsity of spectrographic speech data to formulate the reconstruction problem as a linear program. Each approach is successfully applied in both the Mel-filtered and log Mel-filtered domains. Finally, a statistical approach to Mel-domain mask estimation is proposed, which is used to differentiate between reliable and unreliable time-frequency components. Theory developed for missing feature reconstruction is extended to the application of packet loss concealment during the transmission of speech features over an error-prone channel.;In the case of single-channel speech enhancement, statistical model-based methods are studied. A unified framework is presented for deriving short-time spectral amplitude (STSA) estimators which assume generalized Gamma-distributed speech priors. Additionally, a unified framework is proposed for developing STSA estimators which assume phase equivalence of speech and noise components. Finally, the role of temporal correlation in statistical speech enhancement is explored, resulting in a novel correlation-based STSA estimator. |
