EFFICIENT HIERARCHICAL LABELER ALGORITHM FOR GAUSSIAN LIK ELIHOODS COMPUTATION IN RESOURCE CONSTRAINED SPEECH RECOGNITION SY STEMS

This paper presents a new time/memory-efficient algorithm for the evaluation of state likelihoods in an HMM-based speech recognizer where the states are modeled by Gaussian Mixtures. We first present a fast hierarchical labeling scheme and then an improved version, which is specifically geared toward use in recognizers that use asynchronous sear ch (e.g., stack search) as opposed to synchronous (Viterbi) se arch. The improved labeling algorithm improves the speed of the hierarchical labeler by more than 30%.

[1]  Michael Picheny,et al.  Robust methods for using context-dependent features and models in a continuous speech recognizer , 1994, Proceedings of ICASSP '94. IEEE International Conference on Acoustics, Speech and Signal Processing.

[2]  Lalit R. Bahl,et al.  A tree search strategy for large-vocabulary continuous speech recognition , 1995, 1995 International Conference on Acoustics, Speech, and Signal Processing.

[3]  Michael Picheny,et al.  Performance of the IBM large vocabulary continuous speech recognition system on the ARPA Wall Street Journal task , 1995, 1995 International Conference on Acoustics, Speech, and Signal Processing.

[4]  Lalit R. Bahl,et al.  Partitioning the feature space of a classifier with linear hyperplanes , 1999, IEEE Trans. Speech Audio Process..