Presentation 1999/12/20
An Improvement to 3-D N-best Search Using Path-Distance Based Clustering for Recognizing Multiple Sound Sources
HERACLEOUS Panikos, Satoshi NAKAMURA, Kiyohiro SHIKANO,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) A problem which requires solution in the recognition of distant talking speech is the talker localization. In some approaches the talker is localized by using short- and long-termpower. However, in the case of low SNR the talker localization appears to be difficult. The 3-D Viterbi search based method, proposed by YamadaT. et. al, integrate stalker localization and speech recognition. Although, the method provides high recognition rates its applications are restricted only to the presence of one talker. In previous work we introduced a method able for simultaneous recognition of multiple sound sources. The method is an extension of the 3-D Viterbi search and performs N-best search in a 3-D trellis space composed of input frames, HMM states and direction. This paper describes the improvement of the method by using path-distance based clustering of the hypotheses.
Keyword(in Japanese) (See Japanese page)
Keyword(in English)
Paper # NLC99-107
Date of Issue

Conference Information
Committee NLC
Conference Date 1999/12/20(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 ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) An Improvement to 3-D N-best Search Using Path-Distance Based Clustering for Recognizing Multiple Sound Sources
Sub Title (in English)
Keyword(1)
1st Author's Name HERACLEOUS Panikos
1st Author's Affiliation ()
2nd Author's Name Satoshi NAKAMURA
2nd Author's Affiliation
3rd Author's Name Kiyohiro SHIKANO
3rd Author's Affiliation
Date 1999/12/20
Paper # NLC99-107
Volume (vol) vol.99
Number (no) 523
Page pp.pp.-
#Pages 5
Date of Issue