Presentation | 2008-12-09 Music suppression method for single channel speech mixed with BGM using Bayesian networks Hiroaki ITOU, Takanori NISHINO, Norihide KITAOKA, Kazuya TAKEDA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | A non-parametric stochastic method of the single-channel speech extraction from a mixture of speech and music is proposed. In the proposed method, conventional binary masking based on Bayesian classifier is extended so that the local dependency among time-frequency components of the mixture signal is modeled. A graphical representation of the dependency is introduced as a natural extension of the Bayesian classifier. We show that the proposed approach also extends the VQ based projection method for signal separation. Training scenario and optimal graph topology of the proposed method is exploit through experimental discussions. Finally, the performance of the proposed method is evaluated using six mixtures of speech and pop music. Through the experiments, the effectiveness of the method is clarified by overperforming the conventional method by 4% relative improvement of sound quality. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | single-channel sound source separation / background music suppression / binary masking / Bayesian networks |
Paper # | NLC2008-27,SP2008-82 |
Date of Issue |
Conference Information | |
Committee | NLC |
---|---|
Conference Date | 2008/12/2(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Natural Language Understanding and Models of Communication (NLC) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Music suppression method for single channel speech mixed with BGM using Bayesian networks |
Sub Title (in English) | |
Keyword(1) | single-channel sound source separation |
Keyword(2) | background music suppression |
Keyword(3) | binary masking |
Keyword(4) | Bayesian networks |
1st Author's Name | Hiroaki ITOU |
1st Author's Affiliation | Graduate School of Information Science, Nagoya University() |
2nd Author's Name | Takanori NISHINO |
2nd Author's Affiliation | Graduate School of Information Science, Nagoya University |
3rd Author's Name | Norihide KITAOKA |
3rd Author's Affiliation | Graduate School of Information Science, Nagoya University |
4th Author's Name | Kazuya TAKEDA |
4th Author's Affiliation | Graduate School of Information Science, Nagoya University |
Date | 2008-12-09 |
Paper # | NLC2008-27,SP2008-82 |
Volume (vol) | vol.108 |
Number (no) | 337 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |