Learning transcription errors in speech recognition tasks
US10593320B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | Jan 7, 2018 |
| Grant date | Mar 17, 2020 |
| Priority date | — |
| Expiry date | May 17, 2038 |
Classification
- Technology area (CPC G)Physics
- CPC primaryG10L15/26
- WIPO fieldComputer technology
- WIPO sectorElectrical engineering
Abstract
A mistranscription generated by a speech recognition system is identified. A set of known utterance members is provided for use by a speech recognition system. Each utterance member consists of a respective plural number of words. A received utterance is matched to a first utterance member within the set of known utterance members. The first utterance member is a closest matching utterance member and has a first plural number of words. The matching operation matches fewer than the first plural number of words in the received utterance and the received utterance varies in a first particular manner as compared to a first word in a first slot in the first utterance member. The received utterance is sent to a mistranscription analyzer component which increments evidence that the received utterance is evidence of a mistranscription. Once the incremented evidence for the mistranscription exceeds a threshold, future received utterances containing the mistranscription are treated as though the first word was recognized.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.