Title:
SPEAKER AUTHENTICATION METHOD, SERVER, AND COMPUTER-READABLE STORAGE MEDIUM
Document Type and Number:
WIPO Patent Application WO/2019/179033
Kind Code:
A1
Abstract:
A speaker authentication method, comprising: obtaining voice information of a preset speaker, the content of the voice information being not limited (S301); establishing a 3D convolutional neural network architecture and inputting the voice information of the speaker into the 3D convolutional neural network architecture (S302); creating and storing a voice model of the speaker by means of the 3D convolutional neural network architecture (S303); when a test speech is received, comparing test speech information with the stored voice model of the speaker (S304); and calculating the degree of similarity between the test speech information and the voice model of the speaker; when the degree of similarity is greater than a preset value, the speaker is authenticated successfully; when the degree of similarity is less than the preset value, the speaker fails to be authenticated (S305). The present invention also relates to a server and a computer-readable storage medium.
Inventors:
WANG YIWEN (CN)
WANG JIANZONG (CN)
XIAO JING (CN)
WANG JIANZONG (CN)
XIAO JING (CN)
Application Number:
PCT/CN2018/102203
Publication Date:
September 26, 2019
Filing Date:
August 24, 2018
Export Citation:
Assignee:
PING AN TECH SHENZHEN CO LTD (CN)
International Classes:
G10L17/02
Foreign References:
| CN107464568A | 2017-12-12 | |||
| CN107464568A | 2017-12-12 | |||
| CN107358951A | 2017-11-17 | |||
| CN107404381A | 2017-11-28 | |||
| CN105575388A | 2016-05-11 | |||
| US20170069327A1 | 2017-03-09 | |||
| US20150294670A1 | 2015-10-15 |
Attorney, Agent or Firm:
SHENZHEN WORLD INTELLECTUAL PROPERTY AGENCY (GENERAL PARTNERSHIP) (CN)
Download PDF: