VOICE RECOGNITION PROCESSING METHOD AND APPARATUS - BAIDU ONLINE NETWORK TECHNOLOGY BEIJING CO LTD

Title:

VOICE RECOGNITION PROCESSING METHOD AND APPARATUS

Document Type and Number:

WIPO Patent Application WO/2018/014469

Kind Code:

A1

Abstract:

A voice recognition processing method and apparatus. The method comprises: performing training on a pre-set processing model according to voice sample data of all areas of the whole nation to generate a general Mandarin acoustic model (S110); and respectively performing self-adaptive training on the general Mandarin acoustic model according to voice sample data of each province to generate a Mandarin acoustic model, with a dialectal accent, corresponding to each province (S120). Thereby, based on an accent difference between users from different areas, a Mandarin acoustic model with a dialectal accent is established, and the voice recognition performance is improved.

More Like This:

JP7110057	speech recognition system
JP6812334	Learning data generator for learning autonomous system, learning data generation method for learning autonomous system, program
JP2021105708	NEURAL SPEECH-TO-MEANING

Inventors:

LI XIANGANG (CN)
JIANG BING (CN)

Application Number:

PCT/CN2016/105080

Publication Date:

January 25, 2018

Filing Date:

November 08, 2016

Export Citation:

Click for automatic bibliography generation Help

Assignee:

BAIDU ONLINE NETWORK TECHNOLOGY BEIJING CO LTD (CN)

International Classes:

G10L15/16; G10L15/06

Foreign References:

CN103310788A	2013-09-18
CN104036774A	2014-09-10
CN105632501A	2016-06-01
CN105229725A	2016-01-06
CN105336323A	2016-02-17
US20140288928A1	2014-09-25

Attorney, Agent or Firm:

TSINGYIHUA INTELLECTUAL PROPERTY LLC (CN)

Download PDF:

View/Download PDF PDF Help

Previous Patent: SYSTEM AND METHOD FOR PREPARING POLYMER DISPERSED LIQUID CRYSTAL HOLOGRAPHIC GRATING

Next Patent: METHOD USING SLAG FOR MANUFACTURING MICROCRYSTALLINE MATERIAL