Presentation	2011-07-22 The Forefront of Speech Recognition Algorithms Takaaki HORI,
PDF Download Page	PDF download Page Link
Abstract(in Japanese)	(See Japanese page)
Abstract(in English)	This paper reviews recent advances in speech recognition algorithms. Although there are many algorithms used for speech recognition, decoding algorithms are the focus of this paper, which finds the most likely word sequence for a given speech input. State-of-the-art speech recognition systems employ hidden Markov models for acoustic modeling, word N-gram models for language modeling, and time-synchronous Viterbi beam search for decoding. The decoding algorithm is currently used in the framework of Weighted Finite-State Transducer (WFST)-based speech recognition, which is known as the most efficient approach. This paper mainly describes recent advances in the WFST-based approach, and also some related topics such as consensus decoding, parallel decoding, search optimization, and template-based approach. Finally, some perspectives on the decoder technology are presented.
Keyword(in Japanese)	(See Japanese page)
Keyword(in English)	Speech Recognition / Decoding / Weighted Finite-State Transducer
Paper #	SP2011-47
Date of Issue

Conference Information
Committee	SP
Conference Date	2011/7/14(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)	The Forefront of Speech Recognition Algorithms
Sub Title (in English)
Keyword(1)	Speech Recognition
Keyword(2)	Decoding
Keyword(3)	Weighted Finite-State Transducer
1st Author's Name	Takaaki HORI
1st Author's Affiliation	NTT Communication Science Laboratories()
Date	2011-07-22
Paper #	SP2011-47
Volume (vol)	vol.111
Number (no)	153
Page	pp.pp.-
#Pages	6
Date of Issue