Speaker normalized acoustic modeling based on 3-D viterbi decoding

T. Fukada; Y. Sagisaka

doi:10.1109/ICASSP.1998.674461

Speaker normalized acoustic modeling based on 3-D viterbi decoding

T. Fukada, Y. Sagisaka

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

3 Citations (Scopus)

Abstract

This paper describes a novel method for speaker normalization based on a frequency warping approach to reduce variations due to speaker-induced factors such as the vocal tract length. In our approach, a speaker normalized acoustic model is trained using time-varying (i.e., state, phoneme or word dependent) warping factors, while in the conventional approaches, the frequency warping factor is fixed for each speaker. These time-varying frequency warping factors are determined by a 3-dimensional (i.e., input frames, HMM states and warping factors) Viterbi decoding procedure. Experimental results on Japanese spontaneous speech recognition show that the proposed method yields a 9.7% improvement in speech recognition accuracy compared to the conventional speaker-independent model.

Original language	English
Title of host publication	Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998
Pages	437-440
Number of pages	4
DOIs	https://doi.org/10.1109/ICASSP.1998.674461
Publication status	Published - 1998
Externally published	Yes
Event	1998 23rd IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998 - Seattle, WA, United States Duration: 1998 May 12 → 1998 May 15

Publication series

Name	ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings
Volume	1
ISSN (Print)	1520-6149

Conference

Conference	1998 23rd IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998
Country/Territory	United States
City	Seattle, WA
Period	98/5/12 → 98/5/15

ASJC Scopus subject areas

Software
Signal Processing
Electrical and Electronic Engineering

Access to Document

10.1109/ICASSP.1998.674461

Cite this

Fukada, T., & Sagisaka, Y. (1998). Speaker normalized acoustic modeling based on 3-D viterbi decoding. In Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998 (pp. 437-440). Article 674461 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 1). https://doi.org/10.1109/ICASSP.1998.674461

Speaker normalized acoustic modeling based on 3-D viterbi decoding. / Fukada, T.; Sagisaka, Y.
Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998. 1998. p. 437-440 674461 (ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings; Vol. 1).

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

Fukada, T & Sagisaka, Y 1998, Speaker normalized acoustic modeling based on 3-D viterbi decoding. in Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998., 674461, ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings, vol. 1, pp. 437-440, 1998 23rd IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998, Seattle, WA, United States, 98/5/12. https://doi.org/10.1109/ICASSP.1998.674461

@inproceedings{c8fc1bec418643de9f426df1092c68ce,

title = "Speaker normalized acoustic modeling based on 3-D viterbi decoding",

abstract = "This paper describes a novel method for speaker normalization based on a frequency warping approach to reduce variations due to speaker-induced factors such as the vocal tract length. In our approach, a speaker normalized acoustic model is trained using time-varying (i.e., state, phoneme or word dependent) warping factors, while in the conventional approaches, the frequency warping factor is fixed for each speaker. These time-varying frequency warping factors are determined by a 3-dimensional (i.e., input frames, HMM states and warping factors) Viterbi decoding procedure. Experimental results on Japanese spontaneous speech recognition show that the proposed method yields a 9.7% improvement in speech recognition accuracy compared to the conventional speaker-independent model.",

author = "T. Fukada and Y. Sagisaka",

year = "1998",

doi = "10.1109/ICASSP.1998.674461",

language = "English",

isbn = "0780344286",

series = "ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings",

pages = "437--440",

booktitle = "Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998",

note = "1998 23rd IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998 ; Conference date: 12-05-1998 Through 15-05-1998",

}

TY - GEN

T1 - Speaker normalized acoustic modeling based on 3-D viterbi decoding

AU - Fukada, T.

AU - Sagisaka, Y.

PY - 1998

Y1 - 1998

N2 - This paper describes a novel method for speaker normalization based on a frequency warping approach to reduce variations due to speaker-induced factors such as the vocal tract length. In our approach, a speaker normalized acoustic model is trained using time-varying (i.e., state, phoneme or word dependent) warping factors, while in the conventional approaches, the frequency warping factor is fixed for each speaker. These time-varying frequency warping factors are determined by a 3-dimensional (i.e., input frames, HMM states and warping factors) Viterbi decoding procedure. Experimental results on Japanese spontaneous speech recognition show that the proposed method yields a 9.7% improvement in speech recognition accuracy compared to the conventional speaker-independent model.

AB - This paper describes a novel method for speaker normalization based on a frequency warping approach to reduce variations due to speaker-induced factors such as the vocal tract length. In our approach, a speaker normalized acoustic model is trained using time-varying (i.e., state, phoneme or word dependent) warping factors, while in the conventional approaches, the frequency warping factor is fixed for each speaker. These time-varying frequency warping factors are determined by a 3-dimensional (i.e., input frames, HMM states and warping factors) Viterbi decoding procedure. Experimental results on Japanese spontaneous speech recognition show that the proposed method yields a 9.7% improvement in speech recognition accuracy compared to the conventional speaker-independent model.

UR - http://www.scopus.com/inward/record.url?scp=0031624341&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0031624341&partnerID=8YFLogxK

U2 - 10.1109/ICASSP.1998.674461

DO - 10.1109/ICASSP.1998.674461

M3 - Conference contribution

AN - SCOPUS:0031624341

SN - 0780344286

SN - 9780780344280

T3 - ICASSP, IEEE International Conference on Acoustics, Speech and Signal Processing - Proceedings

SP - 437

EP - 440

BT - Proceedings of the 1998 IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998

T2 - 1998 23rd IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 1998

Y2 - 12 May 1998 through 15 May 1998

ER -

Speaker normalized acoustic modeling based on 3-D viterbi decoding

Abstract

Publication series

Conference

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this