| EPSRC Reference: |
GR/L50280/01 |
| Title: |
ERF: EXPLOITING RECOGNITION FAILURES IN AUTOMATIC RECOGNITION OF DISFLUENT SPEECH |
| Principal Investigator: |
Bard, Dr E |
| Other Investigators: |
|
| Researcher Co-Investigators: |
|
| Project Partners: |
|
| Department: |
Theoretical and Applied Linguistics |
| Organisation: |
University of Edinburgh |
| Scheme: |
Standard Research (Pre-FEC) |
| Starts: |
01 May 1997 |
Ends: |
30 September 2000 |
Value (£): |
291,028
|
| EPSRC Research Topic Classifications: |
| Human Communication in ICT |
|
|
| EPSRC Industrial Sector Classifications: |
| No relevance to Underpinning Sectors |
|
|
| Related Grants: |
|
| Panel History: |
|
|
Summary on Grant Application Form |
|
To become widely usable, automatic speech recognition (ASR) systems must be robustly applicable to spontaneous speech, even when that speech contains disfluencies, false starts, repetitions, corrections and the like. For the most part, current systems are designed to achieve complete recognition of the input, with special strategies deflecting disfluent passages and editing them to produce 'fair copy'. Success rates are low, and disruption spreads from disfluent to surroundinf fluent words. Human listeners are much more successful, often correcting disfluencies without consciously detecting them or being able to recognise many of the words which need to be expunged. This project will modify an existing ASR system to exploit the bases of human performance, principally the disruption by disfluency of the local phrasal context on which human word recognition depends. The process will include coding a corpus of spontaneous speech for prosodic phrasing, aligning the domains within which the HMM and human listeners consolidate their successful paths, developing a criterion for abandoning sub-paths at the end of such domains, and devising correction procedures for the surviving material. ASR results will be evaluated against human performance and in absolute terms.
|
|
Key Findings |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Potential use in non-academic contexts |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
|
Impacts |
|
| Description |
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk |
| Summary |
|
| Date Materialised |
|
|
|
|
Sectors submitted by the Researcher |
|
This information can now be found on Gateway to Research (GtR) http://gtr.rcuk.ac.uk
|
| Project URL: |
|
| Further Information: |
|
| Organisation Website: |
http://www.ed.ac.uk |