Method for converting text into voice of specified style

A voice and style technology, applied in speech synthesis, voice analysis, instruments, etc., can solve problems such as strange changes in intonation, inability to adjust voice intonation, stuttering, etc.

Active Publication Date: 2020-06-26
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF6 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

This "mechanical" method makes computer output audio often produce speech glitches, weird changes in intonation, and even stuttering, and cannot adjust the intonation of speech, emphasizing syllables or emotions

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
  • Method for converting text into voice of specified style
  • Method for converting text into voice of specified style
  • Method for converting text into voice of specified style

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0023] figure 1 It is a flowchart of a specific embodiment of the method for converting text into speech of a specified style in the present invention. Such as figure 1 As shown, the present invention converts text into the specific steps of the method for specifying style speech and includes:

[0024] S101: Obtain training samples:

[0025] Acquire several speech signals of different styles, and divide the speech signals according to the predetermined time interval, record the number of speech signals obtained by segmentation as N, and record the nth speech signal as S n , where n=1,2,...,N, get the speech signal S n the text T n and speech time-frequency image I n , classify the speech signal according to the style, denote the speech signal S n The corresponding style classification label F n . Generally speaking, the style classification label can be set as the serial number of the network, such as F n =1,2,...,K, K represents the number of style types.

[0026] T...

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 method for converting a text into voice of a specified style. The method comprises the steps of: obtaining training samples of different styles ; carrying out character labeling on each voice time-frequency image in the training samples; obtaining a position box sequence and a text character sequence of characters to train a bidirectional long-short-term memory network; constructing a generative adversarial network, taking the voice time-frequency image in the training sample as an original voice time-frequency image, the mask image as a corresponding mask image and the text character sequence as a real text character sequence, and training the generative adversarial network; predicting a text needing to generate a voice signal through a bidirectional long-short-term memory network to obtain a character position box sequence, then obtaining a mask image, cascading the mask image and an original voice time-frequency image of a specified style to serve as inputof a generator module, obtaining a generated voice time-frequency image, and finally obtaining the voice signal through conversion. By adopting the method, conversion from a single text to various specified styles of voice can be realized.

Description

technical field [0001] The invention belongs to the technical field of speech synthesis, and more specifically relates to a method for converting text into speech of a specified style. Background technique [0002] With the deepening of deep learning in the field of speech synthesis, speech generation technology has gradually narrowed the gap between generated speech and human speech. Speech generation usually converts a text into a specific style of speech through a specific synthesis system. This process requires multiple steps, such as front-end text analysis, acoustic modeling, and audio synthesis modules. [0003] The traditional approach to speech generation is concatenative speech synthesis: recording a large number of speech clips from a single speaker, building a large corpus, and then simply selecting from them and synthesizing complete chunks of audio, words, and sentences. This "mechanical" method makes the computer's output audio often produce voice glitches, ...

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
Patent Type & Authority Applications(China)
IPC IPC(8): G10L13/08G10L13/06G10L19/018G10L25/27G10L25/51
CPCG10L13/08G10L13/06G10L13/086G10L19/018G10L25/27G10L25/51G10L2013/083
Inventor 姬艳丽顾天成张浩
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA
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