ALISA: An automatic lightly supervised speech segmentation and alignment tool
- Submitting institution
-
University of Edinburgh
- Unit of assessment
- 11 - Computer Science and Informatics
- Output identifier
- 85693706
- Type
- D - Journal article
- DOI
-
10.1016/j.csl.2015.06.006
- Title of journal
- Computer Speech and Language
- Article number
- -
- First page
- 116
- Volume
- 35
- Issue
- -
- ISSN
- 0885-2308
- Open access status
- Out of scope for open access requirements
- Month of publication
- July
- Year of publication
- 2015
- 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
-
6
- Research group(s)
-
D - Language, Interaction and Robotics
- Citation count
- 14
- Proposed double-weighted
- No
- Reserve for an output with double weighting
- No
- Additional information
- This summarises our development of an efficient, robust algorithm for automatic alignment of roughly-matched speech and text data. This is useful for model training, and as an end in itself. The proposed algorithm has been successfully deployed in generating subtitles for BBC, Channel 4 and Sky TV output, as has been extensively used by Quorate Technology Ltd to train customised speech recognition systems, generating revenues of £100k+ (contact: CEO). The work was cited in a 2019 patent granted to IBM, and was the subject of a keynote talk at the 2018 Baltic HLT conference.
- Author contribution statement
- -
- Non-English
- No
- English abstract
- -