Analysis and Combination of Forward and Backward based Decoders for Improved Speech Transcription

Denis Jouvet 1 Dominique Fohr 1
1 PAROLE - Analysis, perception and recognition of speech
Inria Nancy - Grand Est, LORIA - NLPKD - Department of Natural Language Processing & Knowledge Discovery
Abstract : This paper analysis the behavior of forward and backward-based decoders used for speech transcription. Experiments have showed that backwardbased decoding leads to similar recognition performance as forward-based decoding, which is consistent with the fact that both systems handle similar information through the acoustic, lexical and language models. However, because of heuristics, search algorithms used in decoding explore only a limited portion of the search space. As forward-based and backward-based approaches do not process the speech signal in the same temporal way, they explore different portions of the search space; leading to complementary systems that can be efficiently combined using the ROVER approach. The speech transcription results achieved by combining forward-based and backward-based systems are significantly better than the results obtained by combining the same amount of forward-only or backward-only systems. This confirms the complementary of the forward and backward approaches and thus the usefulness of their combination.
Complete list of metadatas

https://hal.inria.fr/hal-00834296
Contributor : Denis Jouvet <>
Submitted on : Friday, June 14, 2013 - 4:18:11 PM
Last modification on : Tuesday, December 18, 2018 - 4:38:02 PM

Identifiers

  • HAL Id : hal-00834296, version 1

Collections

Citation

Denis Jouvet, Dominique Fohr. Analysis and Combination of Forward and Backward based Decoders for Improved Speech Transcription. TSD - 16th International Conference on Text, Speech and Dialogue - 2013, Sep 2013, Pilsen, Czech Republic. pp.84-91. ⟨hal-00834296⟩

Share

Metrics

Record views

336