Maximum a posteriori adaptation of subspace Gaussian mixture models for cross-lingual speech recognition

Liang Lu; Arnab Ghoshal; Steve Renals

doi:10.1109/ICASSP.2012.6289012

Maximum a posteriori adaptation of subspace Gaussian mixture models for cross-lingual speech recognition

Liang Lu, Arnab Ghoshal, Steve Renals

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

Original language	English
Title of host publication	Acoustics, Speech and Signal Processing (ICASSP), 2012 IEEE International Conference on
Publisher	Institute of Electrical and Electronics Engineers
Pages	4877-4880
Number of pages	4
DOIs	https://doi.org/10.1109/ICASSP.2012.6289012
Publication status	Published - 2012

Keywords / Materials (for Non-textual outputs)

Gaussian processes
hidden Markov models
maximum likelihood estimation
speech recognition
GlobalPhone corpus
HMM-GMM
MAP adaptation approach
SGMM
cross-lingual acoustic modeling
cross-lingual baseline systems
cross-lingual speech recognition
global shared parameter estimation
maximum a posteriori adaptation
out-of-language training data
subspace Gaussian mixture models
target language resources
word error rate reductions
Acoustics
Adaptation models
Covariance matrix
Hidden Markov models
Mathematical model
Speech recognition
Training data
Cross-lingual Speech Recognition
Maximum a Posteriori Adaptation
Subspace Gaussian Mixture Model

Access to Document

10.1109/ICASSP.2012.6289012

Lu Ghoshal renals Maxium A Posteriori Adaptation of Subspace Gaussian Mixture Models for Cross-lingual Speech RecognotionSubmitted manuscript, 251 KB

http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=6289012

Natural Speech Technology
Renals, S. & King, S.
EPSRC
1/05/11 → 31/07/16
Project: Research
SCALE: Speech Communication with Adaptive Learning
King, S. & Renals, S.
EU government bodies
1/01/09 → 31/12/12
Project: Research

Cite this

@inproceedings{5a89cddd7adf4d6e9b9f7843281b1440,

title = "Maximum a posteriori adaptation of subspace Gaussian mixture models for cross-lingual speech recognition",

keywords = "Gaussian processes, hidden Markov models, maximum likelihood estimation, speech recognition, GlobalPhone corpus, HMM-GMM, MAP adaptation approach, SGMM, cross-lingual acoustic modeling, cross-lingual baseline systems, cross-lingual speech recognition, global shared parameter estimation, maximum a posteriori adaptation, out-of-language training data, subspace Gaussian mixture models, target language resources, word error rate reductions, Acoustics, Adaptation models, Covariance matrix, Hidden Markov models, Mathematical model, Speech recognition, Training data, Cross-lingual Speech Recognition, Maximum a Posteriori Adaptation, Subspace Gaussian Mixture Model",

author = "Liang Lu and Arnab Ghoshal and Steve Renals",

year = "2012",

doi = "10.1109/ICASSP.2012.6289012",

language = "English",

pages = "4877--4880",

booktitle = "Acoustics, Speech and Signal Processing (ICASSP), 2012 IEEE International Conference on",

publisher = "Institute of Electrical and Electronics Engineers",

address = "United States",

}

Maximum a posteriori adaptation of subspace Gaussian mixture models for cross-lingual speech recognition. / Lu, Liang; Ghoshal, Arnab; Renals, Steve.
Acoustics, Speech and Signal Processing (ICASSP), 2012 IEEE International Conference on. Institute of Electrical and Electronics Engineers, 2012. p. 4877-4880.

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

TY - GEN

T1 - Maximum a posteriori adaptation of subspace Gaussian mixture models for cross-lingual speech recognition

AU - Lu, Liang

AU - Ghoshal, Arnab

AU - Renals, Steve

PY - 2012

Y1 - 2012

KW - Gaussian processes

KW - hidden Markov models

KW - maximum likelihood estimation

KW - speech recognition

KW - GlobalPhone corpus

KW - HMM-GMM

KW - MAP adaptation approach

KW - SGMM

KW - cross-lingual acoustic modeling

KW - cross-lingual baseline systems

KW - cross-lingual speech recognition

KW - global shared parameter estimation

KW - maximum a posteriori adaptation

KW - out-of-language training data

KW - subspace Gaussian mixture models

KW - target language resources

KW - word error rate reductions

KW - Acoustics

KW - Adaptation models

KW - Covariance matrix

KW - Hidden Markov models

KW - Mathematical model

KW - Speech recognition

KW - Training data

KW - Cross-lingual Speech Recognition

KW - Maximum a Posteriori Adaptation

KW - Subspace Gaussian Mixture Model

U2 - 10.1109/ICASSP.2012.6289012

DO - 10.1109/ICASSP.2012.6289012

M3 - Conference contribution

SP - 4877

EP - 4880

BT - Acoustics, Speech and Signal Processing (ICASSP), 2012 IEEE International Conference on

PB - Institute of Electrical and Electronics Engineers

ER -

Maximum a posteriori adaptation of subspace Gaussian mixture models for cross-lingual speech recognition

Keywords / Materials (for Non-textual outputs)

Access to Document

Projects

Natural Speech Technology

SCALE: Speech Communication with Adaptive Learning

Cite this