Login| Sign Up| Help| Contact|

Patent Searching and Data


Title:
ADVERSARIAL LEARNING-BASED SPEAKER VOICE CONVERSION METHOD AND RELATED DEVICE
Document Type and Number:
WIPO Patent Application WO/2022/142115
Kind Code:
A1
Abstract:
An adversarial learning-based speaker voice conversion method and apparatus, a computer device and a storage medium, the method comprising: preprocessing training data to obtain MFCC features and fundamental frequency features (S11); inputting the MFCC features and the fundamental frequency features into an initial speaker voice conversion model for training (S12); calling an adversarial algorithm to train a content encoder and a content discriminator until a Nash equilibrium state is reached (S13); acquiring a total loss function of a domain discriminator, and detecting whether the total loss function converges (S14); when the detection result is that the total loss function converges, determining a target speaker voice conversion model (S15); acquiring an audio to be converted and a target audio, calling the content encoder to process the audio to be converted to obtain a target content code, and calling an attribute encoder to process the target audio to obtain a target attribute code (S16); and inputting the target content code and the target attribute code into a generator to obtain a converted speaker voice (S17). The described method may improve the efficiency and quality of speaker voice conversion.

Inventors:
LIANG SHUANG (CN)
MIAO CHENFENG (CN)
MA JUN (CN)
WANG SHAOJUN (CN)
Application Number:
PCT/CN2021/096887
Publication Date:
July 07, 2022
Filing Date:
May 28, 2021
Export Citation:
Click for automatic bibliography generation   Help
Assignee:
PING AN TECH SHENZHEN CO LTD (CN)
International Classes:
G10L21/003; G10L21/013
Foreign References:
CN111429893A2020-07-17
CN107293289A2017-10-24
CN111247585A2020-06-05
CN111816156A2020-10-23
CN110060691A2019-07-26
CN111161744A2020-05-15
CN111564160A2020-08-21
US10186251B12019-01-22
US10347241B12019-07-09
Attorney, Agent or Firm:
SHENZHEN SCIENBIZIP INTELLECTUAL PROPERTY AGENCY CO., LTD. (CN)
Download PDF: