Speech enhancement with LSTM recurrent neural networks and its application to noise-robust ASR

Felix Weninger, Hakan Erdogan, Shinji Watanabe, Emmanuel Vincent, Jonathan Le Roux, John R. Hershey, Björn Schuller

Research output: Chapter in Book/Report/Conference proceedingConference contribution

179 Citations (Scopus)

Abstract

We evaluate some recent developments in recurrent neural network (RNN) based speech enhancement in the light of noise-robust automatic speech recognition (ASR). The proposed framework is based on Long Short-Term Memory (LSTM) RNNs which are discriminatively trained according to an optimal speech reconstruction objective. We demonstrate that LSTM speech enhancement, even when used ‘naïvely’ as front-end processing, delivers competitive results on the CHiME-2 speech recognition task. Furthermore, simple, feature-level fusion based extensions to the framework are proposed to improve the integration with the ASR back-end. These yield a best result of 13.76% average word error rate, which is, to our knowledge, the best score to date.

Original languageEnglish
Title of host publicationLatent Variable Analysis and Signal Separation - 12th International Conference, LVA/ICA 2015, Proceedings
PublisherSpringer Verlag
Pages91-99
Number of pages9
Volume9237
ISBN (Print)9783319224817
DOIs
Publication statusPublished - 2015
Externally publishedYes
Event12th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2015 - Liberec, Czech Republic
Duration: 2015 Aug 252015 Aug 28

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9237
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other12th International Conference on Latent Variable Analysis and Signal Separation, LVA/ICA 2015
CountryCzech Republic
CityLiberec
Period15/8/2515/8/28

    Fingerprint

ASJC Scopus subject areas

  • Computer Science(all)
  • Theoretical Computer Science

Cite this

Weninger, F., Erdogan, H., Watanabe, S., Vincent, E., Le Roux, J., Hershey, J. R., & Schuller, B. (2015). Speech enhancement with LSTM recurrent neural networks and its application to noise-robust ASR. In Latent Variable Analysis and Signal Separation - 12th International Conference, LVA/ICA 2015, Proceedings (Vol. 9237, pp. 91-99). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9237). Springer Verlag. https://doi.org/10.1007/978-3-319-22482-4_11