Title:
AUDIO CORPUS SCREENING METHOD AND DEVICE FOR USE IN SPEECH RECOGNITION, AND COMPUTER DEVICE
Document Type and Number:
WIPO Patent Application WO/2020/224119
Kind Code:
A1
Abstract:
Provided in the embodiments of the present application are an audio corpus screening method and device for use in speech recognition, a computer device, and a computer-readable storage medium. The embodiments of the present application relate to the technical field of speech recognition. When screening an audio corpus, the audio corpus is examined via speech activities and segmented to produce original sentences, the audio corpus is annotated with the original sentences serving as units, the audio corpus and an annotated text thereof are used to train a speech recognition model to produce a first speech recognition model, each audio corpus segment is recognized by means of the first speech recognition model to produce a first recognized text of the audio corpus; the first recognized text and the corresponding annotated text are compared to compile statistics of a first word recognition rate of each audio corpus segment, a determination is made on whether the first word recognition rate of each audio corpus segment satisfies a first word recognition rate preset criterion, and, the audio corpus satisfying the first word recognition rate preset criterion and the annotated text of the corpus are stored.
More Like This:
Inventors:
WANG TAO (CN)
Application Number:
PCT/CN2019/103357
Publication Date:
November 12, 2020
Filing Date:
August 29, 2019
Export Citation:
Assignee:
PING AN TECH SHENZHEN CO LTD (CN)
International Classes:
G10L15/06; G10L15/26
Foreign References:
| CN103514369A | 2014-01-15 | |||
| CN108242234A | 2018-07-03 | |||
| CN104318242A | 2015-01-28 | |||
| CN109241997A | 2019-01-18 |
Attorney, Agent or Firm:
SHENZHEN TALENT PATENT SERVICE (CN)
Download PDF: