Many-to-many speaker conversion method based on i vector in non-parallel text condition

A conversion method and speaker technology, applied in speech synthesis, speech analysis, instruments, etc., can solve the problems of unimproved speech similarity, unable to fully express the speaker's personalized characteristics, etc., so as to improve the personality similarity and speech Quality, improving versatility and practicality, and the effect of good application prospects

Active Publication Date: 2019-02-22
NANJING UNIV OF POSTS & TELECOMM
View PDF5 Cites 26 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, since VAWGAN still only uses the speaker's identity label to build a voice conversion system, and the speaker's identity label

Method used

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Image

Smart Image Click on the blue labels to locate them in the text.
Viewing Examples
Smart Image
  • Many-to-many speaker conversion method based on i vector in non-parallel text condition
  • Many-to-many speaker conversion method based on i vector in non-parallel text condition
  • Many-to-many speaker conversion method based on i vector in non-parallel text condition

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0037] Such as figure 1 As shown, the high-quality speech conversion method of the present invention is divided into two parts: the training part is used to obtain the parameters and conversion functions required for speech conversion, and the conversion part is used to convert the source speaker's voice into the target speaker's voice.

[0038] The implementation steps of the training phase are:

[0039] 1.1) Obtain the training corpus of non-parallel text, the training corpus is the corpus of multiple speakers, including the source speaker and the target speaker. The training corpus is taken from the VCC2018 speech corpus. There are 4 male and 4 female speakers in the training set of this corpus, and each speaker has 80 sentence corpus. This method can realize conversion under parallel text, and can also realize conversion under non-parallel text, so these training corpora can also be non-parallel text.

[0040] 1.2) The training corpus uses the WORLD speech analysis / synt...

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

No PUM Login to view more

Abstract

The invention discloses a many-to-many speaker conversion method based on an i vector in a non-parallel text condition. The method comprises a training stage and a conversion stage. A speech conversion system is realized by combining VAWGAN with an i vector, so that the personality similarity and speech quality of the speech after conversion are improved well and a high-quality speech conversion method is realized. In addition, the dependence on the parallel text is eliminated and the speech conversion under the non-parallel text condition is realized. Moreover, no alignment process is neededby the training process. A plurality of source-target speaker conversion systems are integrated into one conversion model, so that multi-speaker-to-multi-speaker conversion is realized. The many-to-many speaker conversion method has the broad application prospects in fields of cross-linguage speech conversion, film dubbing, and speech translation and the like.

Description

technical field [0001] The invention relates to a many-to-many speaker conversion method, in particular to a multi-to-many speaker conversion method based on i vector under non-parallel text conditions. Background technique [0002] Speech conversion is a research branch in the field of speech signal processing, which is carried out and developed on the basis of the research of speech analysis, recognition and synthesis. The goal of voice conversion is to change the voice personality of the source speaker so that it has the voice personality of the target speaker, that is, to make the voice spoken by one person sound like another person's voice after conversion, while preserving semantics . [0003] After years of research on voice conversion technology, many classic conversion methods have emerged. These include most voice conversion methods such as Gaussian Mixed Model (GMM), frequency bending, and deep neural network (DNN). However, most of these speech conversion meth...

Claims

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Application Information

Patent Timeline
no application Login to view more
IPC IPC(8): G10L13/08G10L19/02G10L21/007G10L13/02G10L25/30
CPCG10L13/02G10L13/08G10L19/02G10L21/007G10L25/30
Inventor 李燕萍左宇涛张燕
Owner NANJING UNIV OF POSTS & TELECOMM
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap