CONF Asaei_ICASSP_2011/IDIAP Model-based Compressive Sensing for Multi-party Distant Speech Recognition Asaei, Afsaneh Bourlard, Hervé Cevher, Volkan Model-Based Compressive Sensing Overlapping Speech Sparse Component Analysis Sparse Recovery speech recognition EXTERNAL http://publications.idiap.ch/attachments/papers/2011/Asaei_ICASSP_2011.pdf PUBLIC http://publications.idiap.ch/index.php/publications/showcite/Asaei_Idiap-RR-04-2011 Related documents 2011 IEEE International Conference on Acoustics, Speech and Signal Processing 2011 We leverage the recent algorithmic advances in compressive sensing, and propose a novel source separation algorithm for efficient recovery of convolutive speech mixtures in spectro-temporal domain. Compared to the common sparse component analysis techniques, our approach fully exploits structured sparsity models to obtain substantial improvement over the existing state-of-the-art. We evaluate our method for separation and recognition of a target speaker in a multi-party scenario. Our results provide compelling evidence of the effectiveness of sparse recovery formulations in speech recognition. REPORT Asaei_Idiap-RR-04-2011/IDIAP Model-Based Compressive Sensing for Multi-Party Distant Speech Recognition Asaei, Afsaneh Bourlard, Hervé Cevher, Volkan Model-Based Compressive Sensing Multi-party Speech Recognition Overlapping Speech Sparse Component Analysis Sparse Signal Recovery EXTERNAL http://publications.idiap.ch/attachments/reports/2011/Asaei_Idiap-RR-04-2011.pdf PUBLIC Idiap-RR-04-2011 2011 Idiap March 2011 We leverage the recent algorithmic advances in compressive sensing, and propose a novel source separation algorithm for efficient recovery of convolutive speech mixtures in spectro-temporal domain. Compared to the common sparse component analysis techniques, our approach fully exploits structured sparsity models to obtain substantial improvement over the existing state-of-the-art. We evaluate our method for separation and recognition of a target speaker in a multi-party scenario. Our results provide compelling evidence of the effectiveness of sparse recovery formulations in speech recognition.