Stochastic trajectory modeling for speech recognition
- 17 December 2002
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. i (15206149) , I/57-I/60
- https://doi.org/10.1109/icassp.1994.389356
Abstract
Models observations of phoneme-based speech units as clusters of trajectories in their parameter space. The trajectories are modeled by a mixture of state sequences of multi-variate Gaussian density functions, optimized at the state sequence level. The duration of trajectories are integrated in the modeling. The authors also provide an algorithm for sentence recognition based on the modeling. In an alphabet recognition task the resulting system trained in context-independent mode demonstrated substantially better recognition accuracy, compared to a conventional context-dependent, whole word HMM.Keywords
This publication has 10 references indexed in Scilit:
- Explicit time correlation in hidden Markov models for speech recognitionPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2005
- Noise independent speech recognition for a variety of noise typesPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Phoneme HMMs constrained by frame correlationsPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1993
- Use of temporal correlation between successive frames in a hidden Markov model based speech recognizerPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1993
- Fast algorithms for phone classification and recognition using segment-based modelsIEEE Transactions on Signal Processing, 1992
- A real-time recurrent error propagation network word recognition systemPublished by Institute of Electrical and Electronics Engineers (IEEE) ,1992
- A study on speaker adaptation of the parameters of continuous density hidden Markov modelsIEEE Transactions on Signal Processing, 1991
- A stochastic segment model for phoneme-based continuous speech recognitionIEEE Transactions on Acoustics, Speech, and Signal Processing, 1989
- Speaker-independent isolated word recognition using dynamic features of speech spectrumIEEE Transactions on Acoustics, Speech, and Signal Processing, 1986
- An Algorithm for Vector Quantizer DesignIEEE Transactions on Communications, 1980