An analysis of environment, microphone and data simulation mismatches in robust speech recognition
- Submitting institution
-
The University of Sheffield
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 2440
- Type
- D - Journal article
- DOI
-
10.1016/j.csl.2016.11.005
- Title of journal
- Computer Speech & Language
- Article number
- -
- First page
- 535
- Volume
- 46
- Issue
- -
- ISSN
- 0885-2308
- Open access status
- Compliant
- Month of publication
- December
- Year of publication
- 2016
- URL
-
-
- Supplementary information
-
-
- Request cross-referral to
- -
- Output has been delayed by COVID-19
- No
- COVID-19 affected output statement
- -
- Forensic science
- No
- Criminology
- No
- Interdisciplinary
- No
- Number of additional authors
-
4
- Research group(s)
-
G - Speech and Hearing
- Citation count
- 89
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- A novel analysis of the impact of various forms of ‘mismatch’ that occur in distant microphone speech recognition systems. In particular, it tests assumptions surrounding the use of artificial speech and noise mixing. It directly informed the design of new speech recognition evaluation tasks, CHiME-4 and CHiME-5. CHiME-5 data collection was sponsored by Google ($80,000 contact: Head Google Ireland). Toshiba supported a PhD (£49,000, contact: Group Leader). A collaboration between Sheffield, John Hopkins and Inria commercialised the data leading to £30,000 sales for 15 companies including Toshiba, Hitachi, Sony, Facebook. Non-commercial licences have been distributed to over 250 groups worldwide.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -