A new paradigm for speaker-independent training
- 1 January 1991
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 833-836 vol.2
- https://doi.org/10.1109/icassp.1991.150467
Abstract
A paradigm for speaker-independent (SI) training of hidden Markov models (HMMs) is presented for continuous speech recognition. The method uses a large amount of speech from a few speakers instead of the traditional practice of using a little speech from many speakers. Furthermore, it avoids the common practice of pooling speech data from many speakers prior to training a SI model. The approach has been tested using the BBN Byblos system on the DARPA Resource Management corpus under standard test conditions. With only 12 speakers for training the SI models, recognition performance. comparable to that reported for other systems using 109 training speakers has been achieved. Besides surprisingly good recognition performance, this method offers many practical advantages that have implications for the way speech corpora, are designed and used for training SI models.<>Keywords
This publication has 10 references indexed in Scilit:
- Context-dependent modeling for acoustic-phonetic recognition of continuous speechPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Robust smoothing methods for discrete hidden Markov modelsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- The DARPA 1000-word resource management database for continuous speech recognitionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- On semi-continuous hidden Markov modelingPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Tied mixture continuous parameter modeling for speech recognitionIEEE Transactions on Acoustics, Speech, and Signal Processing, 1990
- Training set issues in SRI's DECIPHER speech recognition systemPublished by Association for Computational Linguistics (ACL) ,1990
- Improved hidden Markov modeling for speaker-independent continuous speech recognitionPublished by Association for Computational Linguistics (ACL) ,1990
- DARPA resource management benchmark test results June 1990Published by Association for Computational Linguistics (ACL) ,1990
- Towards speech recognition without vocabulary-specific trainingPublished by Association for Computational Linguistics (ACL) ,1989
- Unsupervised speaker adaptation based on hierarchical spectral clusteringIEEE Transactions on Acoustics, Speech, and Signal Processing, 1989