Training module for estimating mixture gaussian densities for sp

Details Training module for estimating mixture gaussian densities for sp Training module for estimating mixture gaussian densities for sp

1990-11-15

1993-03-09

Fleming, Michael R.

G10L 900

Patent

active

051931427

ABSTRACT:
A model-training module generates mixture Gaussian density models from speech training data for continuous, or isolated word HMM-based speech recognition systems. Speech feature sequences are labeled into segments of states of speech units using Viterbi-decoding based optimized segmentation algorithm. Each segment is modeled by a Gaussian density, and the parameters are estimated by sample mean and sample covariance. A mixture Gaussian density is generated for each state of each speech unit by merging the Gaussian densities of all the segments with the same corresponding label. The resulting number of mixture components is proportional to the dispension and sample size of the training data. A single, fully merged, Gaussian density is also generated for each state of each speech unit. The covariance matrices of the mixture components are selectively smoothed by a measure of relative sharpness of the Gaussian density. The weights of the mixture components are set uniformly initially, and are reestimated using a segmental-average procedure. The weighting coefficients, together with the Gaussian densities, then become the models of speech units for use in speech recognition.

REFERENCES:
patent: 4032711 (1977-06-01), Sambar
patent: 4241329 (1980-12-01), Bahler et al.
patent: 4741036 (1988-04-01), Bahl et al.
patent: 4833712 (1989-05-01), Bahl et al.
Rabiner et al., "Speaker-Independent Recognition of Isolated Words Using Clustering Techniques", IEEE Trans. On ASSP, vol. ASSP-27, No. 4, Aug. 1979, pp. 336-349.
"Recent Developments in the Application of Hidden Markov Models to Speaker-Independed Isolated Word Recognition".
Biing-Hwang Juang and Lawrence R. Rabiner/Mixture Autoregressive Hidden Markov Models for Speech Signals/Dec. 1985.
Brian Hanson and Hisashi Wakita/Spectral Slope Distance Measures with Linear Prediction Analysis for World Recognition in Noise/Jul. 1987.
Hynek Hermansky, Brian A. Hanson and Hisashi Wakita/Perpetually Based Linear Predictive Analysis of Speech/1985.
C. H. Lee, L. R. Rabiner, R. Pieraccini and J. G. Wilpon/Acoustic Modeling for Large Vocabulary Speech Recognition/1990.
Lawrence R. Rabiner, Jay G. Wilpon and Biing-Hwang Juang/A Segmental k-Means Training Procedure for Connected Word Recognition/AT&T Technical Journal/May/Jun. 1986, vol. 65, Issue 3.

Affiliated with

Also associated with

Rating

Training module for estimating mixture gaussian densities for sp does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Training module for estimating mixture gaussian densities for sp, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Training module for estimating mixture gaussian densities for sp will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-215164