Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
SPEECH RECOGNITION METHOD AND SYSTEM
Document Type and Number:
WIPO Patent Application WO/2021/253779
Kind Code:
A1
Abstract:
Provided are a speech recognition method and system. The method comprises: acquiring voiceprint information of speech information to be recognized (S10); on the basis of the voiceprint information, determining a personalized automatic speech recognition model corresponding to the speech information, wherein the personalized automatic speech recognition model is obtained by means of training on the basis of speech data corresponding to the voiceprint information (S20); and on the basis of the personalized automatic speech recognition model, determining text information corresponding to the speech information (S30). According to the method, the voiceprint information of the speech data used for training the personalized automatic speech recognition model is the same as the voiceprint information of the acquired speech information, and therefore, the speech expression manner of the acquired speech information is the same as a speech expression manner of the speech data used by the personalized automatic speech recognition model, such that the accuracy of speech information recognition can be improved by means of a personalized speech conversion module, and an intention of a user can be better learned, thereby making user usage more convenient.

Inventors:
WEN XIN (CN)
DANG WEIZHEN (CN)
Application Number:
PCT/CN2020/138443
Publication Date:
December 23, 2021
Filing Date:
December 23, 2020
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
SHENZHEN TCL NEW TECH CO LTD (CN)
International Classes:
G10L15/07; G10L15/20; G10L15/26
Foreign References:
CN105096941A2015-11-25
CN103456303A2013-12-18
CN105096940A2015-11-25
CN110634472A2019-12-31
CN111261168A2020-06-09
JP2000214880A2000-08-04
Attorney, Agent or Firm:
JOHNSON INTELLECTUAL PROPERTY AGENCY (SHENZHEN) (CN)
Download PDF: