Presentation | 2004/9/10 DOA estimation and speech signal segregation based on frequency domain binaural model Hidetoshi NAKASHIMA, Yoshifumi CHISAKI, Tsuyoshi USAGAWA, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | As known as a "Cocktail Party Effect", we can communicate others under noisy environments. This effect is based on binaural functions and the human segregates the specific sound by using directional information as a cue of the sound. The computational model for cocktail party effect has been studied, we also proposed it as called "Frequency Domain Binaural Model (FDBM)" which has some characteristics such as less computational load, high segregation quality, and the keep the binaural information of the segregated sound. In this paper, the basic algorithm of FDBM and its performance for segregation obtained by the computer simulations are addressed. According to the evaluation as a speech enhancer, the envelope of the segregated signal is recovered and quite similar to the one of the target signal. On the other hand, more than 90% recognition rates are obtained in speech recognition task, when the azimuth of reception of the target signal and noise differs by 10°. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Binaural Model / Frequency Domain / Interaural Phase Difference / Interaural Level Difference / Head-Related Transfer Function / Sound Segregation / Direction of Arrival Estimation |
Paper # | EA2004-71,SIP2004-75,SIS2004-42 |
Date of Issue |
Conference Information | |
Committee | SIS |
---|---|
Conference Date | 2004/9/10(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 | Smart Info-Media Systems (SIS) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | DOA estimation and speech signal segregation based on frequency domain binaural model |
Sub Title (in English) | |
Keyword(1) | Binaural Model |
Keyword(2) | Frequency Domain |
Keyword(3) | Interaural Phase Difference |
Keyword(4) | Interaural Level Difference |
Keyword(5) | Head-Related Transfer Function |
Keyword(6) | Sound Segregation |
Keyword(7) | Direction of Arrival Estimation |
1st Author's Name | Hidetoshi NAKASHIMA |
1st Author's Affiliation | Kumamoto National College of Technology() |
2nd Author's Name | Yoshifumi CHISAKI |
2nd Author's Affiliation | Faculty of Engineering, Kumamoto University |
3rd Author's Name | Tsuyoshi USAGAWA |
3rd Author's Affiliation | Faculty of Engineering, Kumamoto University |
Date | 2004/9/10 |
Paper # | EA2004-71,SIP2004-75,SIS2004-42 |
Volume (vol) | vol.104 |
Number (no) | 308 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |