A Hybrid MaxEnt/HMM Based ASR System

Yasser Hifny, Steve Renals, Neil D. Lawrence
Proceedings of Interspeech 2005 --- 9th European Conference on Speech Communication and Technology, 2005.

Abstract

The aim of this work is to develop a practical framework, which extends the classical Hidden Markov Model (HMM) for continuous speech recognition based on the Maximum Entropy (MaxEnt) principle. The MaxEnt models can estimate the posterior probabilities directly as with Hybrid NN/HMM connectionist speech recogniton systems. In particular, a new acoustic modelling based on discriminative MaxEnt models is formulated and is being developed to replace the generative Gaussian Mixture Models (GMM) commonly used to model acoustic variability. Initial experimental results using the TIMIT phone task are reported.

Cite this Paper

BibTeX
@InProceedings{Hifny:maxent05, title = {A Hybrid MaxEnt/HMM Based ASR System}, author = {Hifny, Yasser and Renals, Steve and Lawrence, Neil D.}, booktitle = {Proceedings of Interspeech 2005 --- 9th European Conference on Speech Communication and Technology}, year = {2005}, url = {http://inverseprobability.com/publications/hifny-maxent05.html}, abstract = {The aim of this work is to develop a practical framework, which extends the classical Hidden Markov Model (HMM) for continuous speech recognition based on the Maximum Entropy (MaxEnt) principle. The MaxEnt models can estimate the posterior probabilities directly as with Hybrid NN/HMM connectionist speech recogniton systems. In particular, a new acoustic modelling based on discriminative MaxEnt models is formulated and is being developed to replace the generative Gaussian Mixture Models (GMM) commonly used to model acoustic variability. Initial experimental results using the TIMIT phone task are reported.} }
Endnote
%0 Conference Paper %T A Hybrid MaxEnt/HMM Based ASR System %A Yasser Hifny %A Steve Renals %A Neil D. Lawrence %B Proceedings of Interspeech 2005 --- 9th European Conference on Speech Communication and Technology %D 2005 %F Hifny:maxent05 %U http://inverseprobability.com/publications/hifny-maxent05.html %X The aim of this work is to develop a practical framework, which extends the classical Hidden Markov Model (HMM) for continuous speech recognition based on the Maximum Entropy (MaxEnt) principle. The MaxEnt models can estimate the posterior probabilities directly as with Hybrid NN/HMM connectionist speech recogniton systems. In particular, a new acoustic modelling based on discriminative MaxEnt models is formulated and is being developed to replace the generative Gaussian Mixture Models (GMM) commonly used to model acoustic variability. Initial experimental results using the TIMIT phone task are reported.
RIS
TY - CPAPER TI - A Hybrid MaxEnt/HMM Based ASR System AU - Yasser Hifny AU - Steve Renals AU - Neil D. Lawrence BT - Proceedings of Interspeech 2005 --- 9th European Conference on Speech Communication and Technology DA - 2005/09/04 ID - Hifny:maxent05 UR - http://inverseprobability.com/publications/hifny-maxent05.html AB - The aim of this work is to develop a practical framework, which extends the classical Hidden Markov Model (HMM) for continuous speech recognition based on the Maximum Entropy (MaxEnt) principle. The MaxEnt models can estimate the posterior probabilities directly as with Hybrid NN/HMM connectionist speech recogniton systems. In particular, a new acoustic modelling based on discriminative MaxEnt models is formulated and is being developed to replace the generative Gaussian Mixture Models (GMM) commonly used to model acoustic variability. Initial experimental results using the TIMIT phone task are reported. ER -
APA
Hifny, Y., Renals, S. & Lawrence, N.D.. (2005). A Hybrid MaxEnt/HMM Based ASR System. Proceedings of Interspeech 2005 --- 9th European Conference on Speech Communication and Technology Available from http://inverseprobability.com/publications/hifny-maxent05.html.

Related Material