On the Learning Dynamics of Semi-Supervised Training for ASR

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

Abstract

The use of semi-supervised training (SST) has become an increasingly popular way of increasing the performance of ASR acoustic models without the need for further transcribed speech data. However, the performance of the technique can be very sensitive to the quality of the initial ASR system. This paper undertakes a comprehensive study of the improvements gained with respect to variation in the initial systems, the quantity of untranscribed data used, and the learning schedules. We postulate that the reason SST can be effective even when the initial model is poor is because it enables utterance-level information to be propagated to the frame level, and hence hypothesise that the quality of the language model plays a much larger role than the quality of the acoustic model. In experiments on Tagalog data from the IARPA MATERIAL programme, we find that indeed this is the case, and show that with an appropriately chosen recipe it is possible to achieve over 50% relative WER reductions from SST, even when the WER of the initial system is more than 80%.
Original languageEnglish
Title of host publicationProceedings of Interspeech 2021
PublisherInternational Speech Communication Association
Pages716-720
Number of pages5
DOIs
Publication statusPublished - 30 Aug 2021
EventInterspeech 2021: The 22nd Annual Conference of the International Speech Communication Association - Brno, Czech Republic
Duration: 30 Aug 20213 Sep 2021
Conference number: 22
https://www.interspeech2021.org

Publication series

Name
ISSN (Print)1990-9772

Conference

ConferenceInterspeech 2021
Country/TerritoryCzech Republic
CityBrno
Period30/08/213/09/21
Internet address

Keywords

  • speech recognition
  • semi-supervised training

Fingerprint

Dive into the research topics of 'On the Learning Dynamics of Semi-Supervised Training for ASR'. Together they form a unique fingerprint.

Cite this