Valtcho Valtchev

MMIE training of large vocabulary recognition systems

Abstract This paper describes a framework for optimising the structure and parameters of a continuous density HMM-based large vocabulary recognition system using the Maximum Mutual Information Estimation (MMIE) criterion. To reduce the computational complexity of the MMIE training algorithm, confusable segments of speech are identified and stored as word lattices of alternative utterance hypotheses. An iterative mixture splitting procedure is also employed to adjust the number of mixture components in each state during training such that the optimal balance between the number of parameters and the available training data is achieved. Experiments are presented on various test sets from the Wall Street Journal database using up to 66 hours of acoustic training data. These demonstrate that the use of lattices makes MMIE training practicable for very complex recognition systems and large training sets. Furthermore, the experimental results show that MMIE optimisation of system structure and parameters can yield useful increases in recognition accuracy.

The 1994 HTK large vocabulary speech recognition system

This paper describes recent work on the HTK large vocabulary speech recognition system. The system uses tied-state cross-word context-dependent mixture Gaussian HMMs and a dynamic network decoder that can operate in a single pass. In the last year the decoder has been extended to produce word lattices to allow flexible and efficient system development, as well as multi-pass operation for use with computationally expensive acoustic and/or language models. The system vocabulary can now be up to 65 k words, the final acoustic models have been extended to be sensitive to more acoustic context (quinphones), a 4-gram language model has been used and unsupervised incremental speaker adaptation incorporated. The resulting system gave the lowest error rates on both the H1-P0 and H1-C1 hub tasks in the November 1994 ARPA CSR evaluation.

MMI training for continuous phoneme recognition on the TIMIT database

Experiences with a phoneme recognition system for the TIMIT database which uses multiple mixture continuous-density monophone HMMs (hidden Markov models) trained using MMI (maximum mutual information) is reported. A comprehensive set of results are presented comparing the ML (maximum likelihood) and MMI training criteria for both diagonal and full covariance models. These results using simple monophone HMMs show that clear performance gains are achieved by MMI training. These results are comparable with the best reported by others, including those which use context-dependent models. In addition, a number of performance and implementation issues which are crucial to successful MMI training are discussed.<<ETX>>

Lattice-based discriminative training for large vocabulary speech recognition

This paper describes a framework for optimising the parameters of a continuous density HMM-based large vocabulary recognition system using a maximum mutual information estimation (MMIE) criterion. To limit the computational complexity arising from the need to find confusable speech segments in the large search space of alternative utterance hypotheses, word lattices generated from the training data are used. Experiments are presented on the Wall Street journal database using up to 66 hours of training data. These show that lattices combined with an improved estimation algorithm makes MMIE training practicable even for very complex recognition systems and large training sets. Furthermore, experimental results show that MMIE training can yield useful increases in recognition accuracy.

Recurrent input transformations for hidden Markov models

A novel architecture which integrates recurrent input transformation (RITs) and continuous density hidden Markov models (HMMs) is presented. The basic HMM structure is extended to accommodate recurrent neural networks which transform the input observations before they enter the Gaussian output distributions associated with the states of the HMM. During training the parameters of both the HMM and the RIT are simultaneously optimized according to the maximum mutual information (MMI) criterion. Results for the E-set recognition task are presented, demonstrating the ability of RITs to exploit longer-term correlations in the speech signal and to give improved discrimination.<<ETX>>

Discriminative optimisation of large vocabulary recognition systems

Describes a framework for optimising the structure and parameters of a continuous-density HMM-based large-vocabulary speech recognition system using the maximum mutual information estimation (MMIE) criterion. To reduce the computational complexity of the MMIE training algorithm, confusable segments of speech are identified and stored as word lattices of alternative utterance hypotheses. An iterative mixture splitting procedure is also employed to adjust the number of mixture components in each state during training such that the optimal balance between number of parameters and available training data is achieved. Experiments are presented on various test sets from the Wall Street Journal database using the full SI-284 training set. These show that the use of lattices makes MMIE training practicable for very complex recognition systems and large training sets. Furthermore, experimental results demonstrate that MMIE optimisation of system structure and parameters can yield useful increases in recognition accuracy.