A novel speaker adaptation approach for continuous densities HMM's

Eleftherios Frangoulis; Victoria Sgardoni

doi:10.1109/ICASSP.1991.150474

A novel speaker adaptation approach for continuous densities HMM's

Eleftherios Frangoulis, Victoria Sgardoni

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

Abstract

An approach for speaker adaptation aiming to get high recognition performance from an HMM speech recognizer after a short training session with a new speaker is presented. The technique presented exploits the Gaussian multivariate nature of continuous density HMM distributions, to adapt the model parameters. This adaptation technique was applied to a 20-word vocabulary. It was tested on 70 new speakers after a training session of 2 to 5 repetitions of each vocabulary word. The experiments carried out have shown a significant improvement in the recognition performance, even when only two training tokens from the new speaker are used.

Original language	English
Title of host publication	IEEE International Conference on Acoustics, Speech and Signal Processing 1991, ICASSP 91
Place of Publication	Toronto
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	861 - 864
Volume	vol.2
DOIs	https://doi.org/10.1109/ICASSP.1991.150474
Publication status	Published - 14 May 1991

Keywords

hidden Markov models (HMMs)
speech recognition
HMM speech recogniser
continuous densities HMMs

Access to Document

10.1109/ICASSP.1991.150474

Cite this

@inproceedings{65b276e917a547c79fb75d8ba1eafe9c,

title = "A novel speaker adaptation approach for continuous densities HMM's",

abstract = "An approach for speaker adaptation aiming to get high recognition performance from an HMM speech recognizer after a short training session with a new speaker is presented. The technique presented exploits the Gaussian multivariate nature of continuous density HMM distributions, to adapt the model parameters. This adaptation technique was applied to a 20-word vocabulary. It was tested on 70 new speakers after a training session of 2 to 5 repetitions of each vocabulary word. The experiments carried out have shown a significant improvement in the recognition performance, even when only two training tokens from the new speaker are used.",

keywords = "hidden Markov models (HMMs), speech recognition, HMM speech recogniser, continuous densities HMMs",

author = "Eleftherios Frangoulis and Victoria Sgardoni",

year = "1991",

month = may,

day = "14",

doi = "10.1109/ICASSP.1991.150474",

language = "English",

volume = "vol.2",

pages = "861 -- 864",

booktitle = "IEEE International Conference on Acoustics, Speech and Signal Processing 1991, ICASSP 91",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

address = "United States",

}

A novel speaker adaptation approach for continuous densities HMM's. / Frangoulis, Eleftherios; Sgardoni, Victoria.

IEEE International Conference on Acoustics, Speech and Signal Processing 1991, ICASSP 91. Vol. vol.2 Toronto : Institute of Electrical and Electronics Engineers (IEEE), 1991. p. 861 - 864.

Research output: Chapter in Book/Report/Conference proceeding › Conference Contribution (Conference Proceeding)

TY - GEN

T1 - A novel speaker adaptation approach for continuous densities HMM's

AU - Frangoulis, Eleftherios

AU - Sgardoni, Victoria

PY - 1991/5/14

Y1 - 1991/5/14

N2 - An approach for speaker adaptation aiming to get high recognition performance from an HMM speech recognizer after a short training session with a new speaker is presented. The technique presented exploits the Gaussian multivariate nature of continuous density HMM distributions, to adapt the model parameters. This adaptation technique was applied to a 20-word vocabulary. It was tested on 70 new speakers after a training session of 2 to 5 repetitions of each vocabulary word. The experiments carried out have shown a significant improvement in the recognition performance, even when only two training tokens from the new speaker are used.

AB - An approach for speaker adaptation aiming to get high recognition performance from an HMM speech recognizer after a short training session with a new speaker is presented. The technique presented exploits the Gaussian multivariate nature of continuous density HMM distributions, to adapt the model parameters. This adaptation technique was applied to a 20-word vocabulary. It was tested on 70 new speakers after a training session of 2 to 5 repetitions of each vocabulary word. The experiments carried out have shown a significant improvement in the recognition performance, even when only two training tokens from the new speaker are used.

KW - hidden Markov models (HMMs)

KW - speech recognition

KW - HMM speech recogniser

KW - continuous densities HMMs

U2 - 10.1109/ICASSP.1991.150474

DO - 10.1109/ICASSP.1991.150474

M3 - Conference Contribution (Conference Proceeding)

VL - vol.2

SP - 861

EP - 864

BT - IEEE International Conference on Acoustics, Speech and Signal Processing 1991, ICASSP 91

PB - Institute of Electrical and Electronics Engineers (IEEE)

CY - Toronto

ER -

A novel speaker adaptation approach for continuous densities HMM's

Abstract

Keywords

Access to Document

Fingerprint

Cite this