APPROXIMATE NEAREST-SUBSPACE REPRESENTATIONS FOR SOUND MIXTURES
Innovative Representations of Audio
Presented by: Paris Smaragdis, Author(s): Paris Smaragdis, University of Illinois Urbana-Champaign, United States
In this paper we present a novel approach to describe sound mixtures which is based on a geometric viewpoint. In this approach we extend the idea of a nearest-neighbor representation to address the case of superimposed sources. We show that in order to account for mixing effects we need to perform a search for nearest-subspaces, as opposed to nearest-neighbors. In order to reduce the excessive computational complexity of this search we present an efficient algorithm to solve this problem which amounts to a sparse coding approach. We demonstrate the efficacy of this algorithm by using it to separate mixtures of speech.
Lecture Information
| Recorded: | 2011-05-26 11:10 - 11:30, Club D |
|---|---|
| Added: | 18. 6. 2011 23:30 |
| Number of views: | 35 |
| Video resolution: | 1024x576 px, 512x288 px |
| Video length: | 0:21:12 |
| Audio track: | MP3 [7.17 MB], 0:21:12 |
Related Lectures
0:24:35
LEARNING A BETTER REPRESENTATION OF SPEECH SOUND WAVES USING RESTRICTED BOLTZMANN MACHINES
Innovative Representations of Audio
Added: 15. 6. 2011 08:59
0:18:47
JOINT SOURCE-FILTER MODELING USING FLEXIBLE BASIS FUNCTIONS
Innovative Representations of Audio
Added: 22. 6. 2011 03:55
Comments