Abstract | ||
---|---|---|
We introduce a new technique for using the speech of multiple reference speakers as a basis for speaker adaptation in large vocabulary continuous speech recognition. In contrast to other methods that use a pooled reference model, this technique normalizes the training speech from multiple reference speakers to a single common feature space before pooling it. The normalized and pooled speech can then be treated as if it came from a single reference speaker for training the reference hidden Markov model (HMM). Our usual probabilistic spectrum transformation can be applied to the reference HMM to model a new (target) speaker. In this paper, we describe our baseline (single reference speaker) speaker-adaptation system and give current performance results from a recent formal evaluation of the system. We also describe our proposal for adapting from multiple reference speakers and report on recent preliminary experimental results in support of the proposed technique. |
Year | DOI | Venue |
---|---|---|
1989 | 10.3115/1075434.1075476 | HLT |
Keywords | DocType | ISBN |
markov model,multiple reference speaker,reference hmm,training speech,single reference speaker,pooled reference model,large vocabulary continuous speech,pooled speech,new technique,speaker adaptation,hidden markov model,mathematical models,adaptation,feature space,markov processes,spectrum,reference model,transformations,spectra,speech recognition | Conference | 1-55860-112-0 |
Citations | PageRank | References |
2 | 2.32 | 3 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Francis Kubala | 1 | 410 | 99.88 |
Richard Schwartz | 2 | 160 | 98.45 |
Chris Barry | 3 | 20 | 9.70 |