A compact model for speaker-adaptive training
Top Cited Papers
- 24 December 2002
- proceedings article
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 2, 1137-1140
- https://doi.org/10.1109/icslp.1996.607807
Abstract
In this work we formulate a novel approach to estimating the pa- rameters of continuous density HMMs for speaker-independent (SI) continuous speech recognition. It is motivated by the fact that vari- ability in SI acoustic models is attributed to both phonetic variation and variation among the speakers of the training population, that is independent of the information content of the speech signal. These two variation sources are decoupled and the proposed method jointly annihilates the inter-speaker variation and estimates the HMM pa- rameters of the SI acoustic models. We compare the proposed training algorithm to the common SI training paradigm within the context of supervised adaptation. We show that the proposed acoustic models are more efficiently adapted to the test speakers, thus achieving significant overall word error rate reductions of 19% and 25% for 20K and 05K vocabulary tasks respectively.Keywords
This publication has 8 references indexed in Scilit:
- A parametric approach to vocal tract length normalizationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Adaptation to new microphones using tied-mixture normalizationPublished by Institute of Electrical and Electronics Engineers (IEEE) ,2002
- Signal bias removal by maximum likelihood estimation for robust telephone speech recognitionIEEE Transactions on Speech and Audio Processing, 1996
- Flexible speaker adaptation for large vocabulary speech recognitionPublished by International Speech Communication Association ,1995
- Stochastic matching for robust speech recognitionIEEE Signal Processing Letters, 1994
- An acoustic-phonetic-based speaker adaptation technique for improving speaker-independent continuous speech recognitionIEEE Transactions on Speech and Audio Processing, 1994
- The metamorphic algorithm: a speaker mapping approach to data augmentationIEEE Transactions on Speech and Audio Processing, 1994
- A Maximization Technique Occurring in the Statistical Analysis of Probabilistic Functions of Markov ChainsThe Annals of Mathematical Statistics, 1970