Timestamp-aligning and keyword-biasing end-to-end ASR front-end for a KWS system
Timestamp-aligning and keyword-biasing end-to-end ASR front-end for a KWS system
About this item
Full title
Author / Creator
Publisher
Cham: Springer International Publishing
Journal title
Language
English
Formats
Publication information
Publisher
Cham: Springer International Publishing
Subjects
More information
Scope and Contents
Contents
Many end-to-end approaches have been proposed to detect predefined keywords. For scenarios of multi-keywords, there are still two bottlenecks that need to be resolved: (1) the distribution of important data that contains keyword(s) is sparse, and (2) the timestamps of the detected keywords are inaccurate. In this paper, to alleviate the first issue...
Alternative Titles
Full title
Timestamp-aligning and keyword-biasing end-to-end ASR front-end for a KWS system
Authors, Artists and Contributors
Author / Creator
Identifiers
Primary Identifiers
Record Identifier
TN_cdi_doaj_primary_oai_doaj_org_article_03c17f151a4b4ebea782da8ab0e5c2bf
Permalink
https://devfeature-collection.sl.nsw.gov.au/record/TN_cdi_doaj_primary_oai_doaj_org_article_03c17f151a4b4ebea782da8ab0e5c2bf
Other Identifiers
ISSN
1687-4722,1687-4714
E-ISSN
1687-4722
DOI
10.1186/s13636-021-00212-9
