SuperLectures.com

MODEL-BASED COMPRESSIVE SENSING FOR MULTI-PARTY DISTANT SPEECH RECOGNITION

Robust ASR

Full Paper at IEEE Xplore

Presented by: Afsaneh Asaei, Author(s): Afsaneh Asaei, Hervé Bourlard, Volkan Cevher, Idiap Research Institute / Ecole Polytechnique Federale de Lausanne, Switzerland

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.


  Speech Transcript

|

  Slides

Enlarge the slide | Show all slides in a pop-up window

0:00:22

  1. slide

0:01:06

  2. slide

0:01:31

  3. slide

0:01:57

  4. slide

0:02:54

  5. slide

0:04:07

  6. slide

0:05:32

  7. slide

0:05:45

  8. slide

0:07:10

  9. slide

0:08:28

 10. slide

0:09:47

 11. slide

0:10:45

 12. slide

0:11:52

 13. slide

0:12:34

 14. slide

0:13:22

 15. slide

0:14:10

 16. slide

  Comments

Please sign in to post your comment!

  Lecture Information

Recorded: 2011-05-26 17:35 - 17:55, Panorama
Added: 15. 6. 2011 19:18
Number of views: 43
Video resolution: 1024x576 px, 512x288 px
Video length: 0:18:54
Audio track: MP3 [6.00 MB], 0:18:54