Context dependent vector quantization for continuous speech recognition
- 1 January 1993
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2 (15206149) , 632-635 vol.2
- https://doi.org/10.1109/icassp.1993.319390
Abstract
The authors present a method for designing a vector quantizer for speech recognition that uses decision networks constructed by examining the phonetic context to obtain models for classes in the quantizer. Diagonal Gaussian models are constructed for the vector quantizer classes at each terminal node of the network and are used to label speech parameter vectors during recognition. Experimental results indicate that this method leads to superior vector quantizers for continuous speech.Keywords
This publication has 5 references indexed in Scilit:
- Context-dependent modeling for acoustic-phonetic recognition of continuous speechPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Large vocabulary natural language continuous speech recognitionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2003
- Allophone clustering for continuous speech recognitionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Modelling context dependency in acoustic-phonetic and lexical representationsPublished by Association for Computational Linguistics (ACL) ,1991
- Decision trees for phonological rules in continuous speechPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1991