Presentation 1993/6/18
A Frame Synchronous Word Spotting Method using a-posteriori Probabilities
Akihiro Imamura, Mikio Kitai,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) The high rate of False alarm is a serious problem in the Keyword Spotting methods based on the continuous Viterbi decoding using HMMS.To overcome this problem,in the paper,we propose a frame- synchronous Keyword Spotting method which extracts candidates for the word sequences accepted by a syntactic automaton with garbage words and keywords.The HMMs for the garbage words are estimated from the non-speech periods and the utterances for non-keywords which will be appeared in spontaneous input speech.Comparing the a- posteriori probabilities obtained by using the One-Pass search algorithm,candidates are extracted accurately.The detail of method and experimental results which compare garbage HMMs are described.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Word-Spotting / Telephone speech recognition / HMM / a- posteriori probability
Paper # SP93-31
Date of Issue

Conference Information
Committee SP
Conference Date 1993/6/18(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 Speech (SP)
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Frame Synchronous Word Spotting Method using a-posteriori Probabilities
Sub Title (in English)
Keyword(1) Word-Spotting
Keyword(2) Telephone speech recognition
Keyword(3) HMM
Keyword(4) a- posteriori probability
1st Author's Name Akihiro Imamura
1st Author's Affiliation NTT Human Interface Laboratory()
2nd Author's Name Mikio Kitai
2nd Author's Affiliation NTT Human Interface Laboratory
Date 1993/6/18
Paper # SP93-31
Volume (vol) vol.93
Number (no) 88
Page pp.pp.-
#Pages 8
Date of Issue