Voice conversion method

A technology of speech conversion and speech signal, applied in speech synthesis, speech analysis, instruments, etc., can solve the problem of long time consumption, and achieve the effect of short time consumption, simple calculation and low energy consumption

Inactive Publication Date: 2015-12-30
CHANGZHOU INST OF TECH
View PDF6 Cites 6 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

To a certain extent, it solves the defect that other algorithms take a long time and are not suitable for devices with small memory

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
  • Voice conversion method
  • Voice conversion method
  • Voice conversion method

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0034] The present invention will be described in detail below in conjunction with the accompanying drawings and specific embodiments.

[0035] The present invention extracts the STRAIGHT spectrum of the voice signal through the STRAIGHT model, performs approximate processing on the STRAIGHT spectrum through the mixed Gaussian model, obtains the weight of the characteristic parameters, uses the characteristic parameters to establish the conditional distribution function between the source and the target voice, and uses the conditional distribution function when converting and source voice feature parameters to realize voice conversion. Specifically include the following steps:

[0036] (1) adopting the speech analysis model to analyze the original speech signal;

[0037] (2) Fix the Mel frequency, and sample the analyzed STRAIGHT spectrum as the sampling frequency;

[0038] (3) Calculate the standard deviation from the frequency spectrum obtained by sampling as the bandwidth...

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 voice conversion method. Firstly, a feature extraction algorithm is used for reducing dimensionality of a model parameter; Mel frequency is set in advance to be used as sampling frequency, and the STRAIGHT spectrum of a voice signal is sampled, so that a Mel frequency logarithmic spectrum is obtained; a standard deviation is calculated by using MFLS; near perfect reconstruction is carried out on the STRAIGHT spectrum through a Gaussian mixture model by using the Mel fixed frequency, the MFLS and the standard deviation, so that feature parameters are obtained; secondly, the method disclosed by the invention is different from a conventional method for estimating a statistic model from training data; by establishing a condition histogram and approximately evaluating conditional distribution of target voices, voice conversion is realized; the characteristics of limited own resources of an embedded system, a mobile device and the like are fully combined, so that high efficiency, short consuming time, simple calculation and low energy consumption are realized.

Description

technical field [0001] The invention relates to voice conversion technology, in particular to a voice conversion method combining voice recognition and voice synthesis technology. Background technique [0002] Speech conversion technology is an emerging research branch in the field of speech signal processing in recent years, covering the fields of speech recognition and speech synthesis. A person's voice personality characteristics, so that his (or her) speech is perceived by the listener as the speech of another specific speaker (called the target speaker). The main tasks of speech conversion include extracting characteristic parameters representing the speaker's personality and performing mathematical transformation, and then reconstructing the transformed parameters into speech. In this process, it is necessary to maintain the auditory quality of the reconstructed voice, but also to take into account whether the converted personality characteristics are accurate. [00...

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): G10L13/08G10L17/04
Inventor 鲍静益徐宁
Owner CHANGZHOU INST OF TECH
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