DMCA
Tandem Acoustic Modeling In Large-Vocabulary Recognition (2001)
Cached
Download Links
| Venue: | in Proc. ICASSP-2001 |
| Citations: | 44 - 2 self |
BibTeX
@INPROCEEDINGS{Ellis01tandemacoustic,
author = {Daniel Ellis and Rita Singh and Sunil Sivadas},
title = {Tandem Acoustic Modeling In Large-Vocabulary Recognition},
booktitle = {in Proc. ICASSP-2001},
year = {2001},
pages = {517--520}
}
Share
 |
 |
 |
 |
||
OpenURL
Abstract
In the tandem approach to modeling the acoustic signal, a neural-net preprocessor is first discriminatively trained to estimate posterior probabilities across a phone set. These are then used as feature inputs for a conventional hidden Markov model (HMM) based speech recognizer, which relearns the associations to subword units. In this paper, we apply the tandem approach to the data provided for the first Speech in Noisy Environments (SPINE1) evaluation conducted by the Naval Research Laboratory (NRL) in August 2000. In our previous experience with the ETSI Aurora noisy digits (a small-vocabulary, high-noise task) the tandem approach achieved error-rate reductions of over 50% relative to the HMM baseline. For SPINE1, a larger task involving more spontaneous speech, we find that, when context-independent models are used, the tandem features continue to result in large reductions in word-error rates relative to those achieved by systems using standard MFC or PLP features. However, these ...
Keyphrases
tandem approach tandem acoustic modeling large-vocabulary recognition acoustic signal naval research laboratory noisy environment large reduction hmm baseline speech recognizer tandem feature feature input high-noise task posterior probability standard mfc phone set spontaneous speech context-independent model first speech conventional hidden markov model previous experience word-error rate neural-net preprocessor plp feature error-rate reduction etsi aurora noisy digit
