An optimized multi-duration HMM for spontaneous speech recognition

Yuichi Ohkawa; Akihiro Yoshida; Motoyuki Suzuki; Akinori Ito; Shozo Makino

An optimized multi-duration HMM for spontaneous speech recognition

Yuichi Ohkawa, Akihiro Yoshida, Motoyuki Suzuki, Akinori Ito, Shozo Makino

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

In spontaneous speech, various speech style and speed changes can be observed, which are known to degrade speech recognition accuracy. In this paper, we describe an optimized multi-duration HMM (OMD). An OMD is a kind of multi-path HMM with at most two parallel paths. Each path is trained using speech samples with short or long phoneme duration. The thresholds to divide samples of phonemes are determined through phoneme recognition experiment. Not only the thresholds but also topologies of HMM are determined using the recognition result. Next, we parallelize OMD model with ordinary HMM trained by spontaneous speech and HMM trained by read speech in parallel. Using this 'all-parallel' model, 19.3% reduction of word error rate was obtained compared with the ordinary HMM trained with spontaneous speech.

Original language	English
Title of host publication	EUROSPEECH 2003 - 8th European Conference on Speech Communication and Technology
Publisher	International Speech Communication Association
Pages	485-488
Number of pages	4
Publication status	Published - 2003
Event	8th European Conference on Speech Communication and Technology, EUROSPEECH 2003 - Geneva, Switzerland Duration: 2003 Sep 1 → 2003 Sep 4

Other

Other	8th European Conference on Speech Communication and Technology, EUROSPEECH 2003
Country/Territory	Switzerland
City	Geneva
Period	03/9/1 → 03/9/4

ASJC Scopus subject areas

Computer Science Applications
Software
Linguistics and Language
Communication

Cite this

An optimized multi-duration HMM for spontaneous speech recognition. / Ohkawa, Yuichi; Yoshida, Akihiro; Suzuki, Motoyuki; Ito, Akinori; Makino, Shozo.

EUROSPEECH 2003 - 8th European Conference on Speech Communication and Technology. International Speech Communication Association, 2003. p. 485-488.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Ohkawa, Y, Yoshida, A, Suzuki, M, Ito, A & Makino, S 2003, An optimized multi-duration HMM for spontaneous speech recognition. in EUROSPEECH 2003 - 8th European Conference on Speech Communication and Technology. International Speech Communication Association, pp. 485-488, 8th European Conference on Speech Communication and Technology, EUROSPEECH 2003, Geneva, Switzerland, 03/9/1.

@inproceedings{56244ad281d8439fbb99096377c4a629,

title = "An optimized multi-duration HMM for spontaneous speech recognition",

abstract = "In spontaneous speech, various speech style and speed changes can be observed, which are known to degrade speech recognition accuracy. In this paper, we describe an optimized multi-duration HMM (OMD). An OMD is a kind of multi-path HMM with at most two parallel paths. Each path is trained using speech samples with short or long phoneme duration. The thresholds to divide samples of phonemes are determined through phoneme recognition experiment. Not only the thresholds but also topologies of HMM are determined using the recognition result. Next, we parallelize OMD model with ordinary HMM trained by spontaneous speech and HMM trained by read speech in parallel. Using this 'all-parallel' model, 19.3% reduction of word error rate was obtained compared with the ordinary HMM trained with spontaneous speech.",

author = "Yuichi Ohkawa and Akihiro Yoshida and Motoyuki Suzuki and Akinori Ito and Shozo Makino",

year = "2003",

language = "English",

pages = "485--488",

booktitle = "EUROSPEECH 2003 - 8th European Conference on Speech Communication and Technology",

publisher = "International Speech Communication Association",

note = "8th European Conference on Speech Communication and Technology, EUROSPEECH 2003 ; Conference date: 01-09-2003 Through 04-09-2003",

}

TY - GEN

T1 - An optimized multi-duration HMM for spontaneous speech recognition

AU - Ohkawa, Yuichi

AU - Yoshida, Akihiro

AU - Suzuki, Motoyuki

AU - Ito, Akinori

AU - Makino, Shozo

PY - 2003

Y1 - 2003

N2 - In spontaneous speech, various speech style and speed changes can be observed, which are known to degrade speech recognition accuracy. In this paper, we describe an optimized multi-duration HMM (OMD). An OMD is a kind of multi-path HMM with at most two parallel paths. Each path is trained using speech samples with short or long phoneme duration. The thresholds to divide samples of phonemes are determined through phoneme recognition experiment. Not only the thresholds but also topologies of HMM are determined using the recognition result. Next, we parallelize OMD model with ordinary HMM trained by spontaneous speech and HMM trained by read speech in parallel. Using this 'all-parallel' model, 19.3% reduction of word error rate was obtained compared with the ordinary HMM trained with spontaneous speech.

AB - In spontaneous speech, various speech style and speed changes can be observed, which are known to degrade speech recognition accuracy. In this paper, we describe an optimized multi-duration HMM (OMD). An OMD is a kind of multi-path HMM with at most two parallel paths. Each path is trained using speech samples with short or long phoneme duration. The thresholds to divide samples of phonemes are determined through phoneme recognition experiment. Not only the thresholds but also topologies of HMM are determined using the recognition result. Next, we parallelize OMD model with ordinary HMM trained by spontaneous speech and HMM trained by read speech in parallel. Using this 'all-parallel' model, 19.3% reduction of word error rate was obtained compared with the ordinary HMM trained with spontaneous speech.

UR - http://www.scopus.com/inward/record.url?scp=85009227728&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85009227728&partnerID=8YFLogxK

M3 - Conference contribution

AN - SCOPUS:85009227728

SP - 485

EP - 488

BT - EUROSPEECH 2003 - 8th European Conference on Speech Communication and Technology

PB - International Speech Communication Association

T2 - 8th European Conference on Speech Communication and Technology, EUROSPEECH 2003

Y2 - 1 September 2003 through 4 September 2003

ER -

An optimized multi-duration HMM for spontaneous speech recognition

Abstract

Other

ASJC Scopus subject areas

Other files and links

Fingerprint

Cite this