Auxiliary loss function for target speech extraction and recognition with weak supervision based on speaker characteristics

Katerina Zmolikova*, Marc Delcroix, Desh Raj, Shinji Watanabe, Jan Černocký

*Corresponding author for this work

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

Abstract

Automatic speech recognition systems deteriorate in presence of overlapped speech. A popular approach to alleviate this is target speech extraction. The extraction system is usually trained with a loss function measuring the discrepancy between the estimated and the reference target speech. This often leads to distortions to the target signal which is detrimental to the recognition accuracy. Additionally, it is necessary to have the strong supervision provided by parallel data consisting of speech mixtures and single-speaker signals. We propose an auxiliary loss function for retraining the target speech extraction. It is composed of two parts: first, a speaker identity loss, forcing the estimated speech to have correct speaker characteristics, and second, a mixture consistency loss, making the extracted sources sum back to the original mixture. The only supervision required for the proposed loss is speaker characteristics obtained from several segments spoken by the target speaker. Such weak supervision makes the loss suitable for adapting the system directly on real recordings. We show that the proposed loss yields signals more suitable for speech recognition and further, we can gain additional improvements by adaptation to target data. Overall, we can reduce the word error rate on LibriCSS dataset from 27.4% to 24.0%.

Original languageEnglish
Title of host publication22nd Annual Conference of the International Speech Communication Association, INTERSPEECH 2021
PublisherInternational Speech Communication Association
Pages4156-4160
Number of pages5
ISBN (Electronic)9781713836902
DOIs
Publication statusPublished - 2021
Externally publishedYes
Event22nd Annual Conference of the International Speech Communication Association, INTERSPEECH 2021 - Brno, Czech Republic
Duration: 2021 Aug 302021 Sep 3

Publication series

NameProceedings of the Annual Conference of the International Speech Communication Association, INTERSPEECH
Volume6
ISSN (Print)2308-457X
ISSN (Electronic)1990-9772

Conference

Conference22nd Annual Conference of the International Speech Communication Association, INTERSPEECH 2021
Country/TerritoryCzech Republic
CityBrno
Period21/8/3021/9/3

Keywords

  • Long recordings
  • SpeakerBeam
  • Target speech extraction
  • Weakly supervised loss

ASJC Scopus subject areas

  • Language and Linguistics
  • Human-Computer Interaction
  • Signal Processing
  • Software
  • Modelling and Simulation

Fingerprint

Dive into the research topics of 'Auxiliary loss function for target speech extraction and recognition with weak supervision based on speaker characteristics'. Together they form a unique fingerprint.

Cite this