Low-rate voice coding method based on depth self-coding machine

A voice coding, low-rate technology, applied in speech analysis, instruments, etc., can solve the problems of lower coding rate, high voice coding quality, etc., and achieve the effect of digitization and compression coding

Pending Publication Date: 2020-06-05
NAT UNIV OF DEFENSE TECH
View PDF3 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a novel low-rate speech encoding method based on a deep autoencoder to solve the technical problem that it is difficult to further reduce the encoding rate while maintaining a higher speech encoding quality in the prior art

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
  • Low-rate voice coding method based on depth self-coding machine
  • Low-rate voice coding method based on depth self-coding machine
  • Low-rate voice coding method based on depth self-coding machine

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0047] This embodiment provides a low-rate speech coding method based on a deep autoencoder, as shown in Figure 1, including:

[0048] Step 1, input the original speech signal s(n), and process the original speech signal s(n) into frames to obtain each frame of speech signal s m ; Wherein, n represents the time subscript, 0≤n≤L-1, L represents the frame length, m represents the subscript of each frame of speech signal, m=1,2,...,M, M represents the total number of speech frame number;

[0049] Wherein, by formula (1), the original speech signal s(n) is subjected to frame processing to obtain each frame of speech signal s m ;

[0050] the s m (n)=s m (mR+n)ω(n) (1)

[0051] In formula (1), R represents the frame shift, and ω(n) represents the Hamming window.

[0052] In this embodiment, the length of each frame of voice signal is 20-30 ms, and the length of each frame of voice signal s m Can be expressed as: s m =[s m (0),s m (1),...,s m (L-1)] T ;

[0053] Step 2,...

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 provides a low-rate voice coding method based on a depth self-coding machine. The method comprises the steps: 1, inputting an original voice signal s (n), carrying out the framing of theoriginal voice signal s (n), and obtaining each frame of voice signal sm; step 2, taking a logarithm amplitude spectrum ym of each frame of voice signal sm after framing processing; 3, constructing adeep neural network model, and training the constructed deep neural network model; and step 4, inputting the logarithm amplitude spectrum ym of each frame of voice signal into the trained deep neuralnetwork model, step 5, obtaining each frame of reconstructed voice signal; and carrying out overlapping operation on each reconstructed frame of voice signal to obtain voice codes, and outputting thevoice codes. According to the invention, a data driving mode is adopted to automatically learn from the voice signals to obtain feature parameters capable of carrying out quantization coding, and digitization and compression coding of the voice signals are realized by carrying out efficient quantization on the feature parameters.

Description

technical field [0001] The invention belongs to the technical field of low-rate vocoders in speech coding, and in particular relates to a low-rate speech coding method based on a deep autoencoder. Background technique [0002] Voice communication is the most natural and convenient means for human beings to communicate with each other. With the rapid development of the mobile Internet, although the volume of data communication business has surpassed the traditional voice communication business, the basic position of voice communication will not change for a long time. Speech coding, which aims at efficiently compressing speech signals by means of digital signal processing to meet the needs of limited communication bandwidth, is one of the core and key technologies of speech communication. With years of in-depth research, many successful speech coding models have been proposed and a series of speech compression coding standards have been formulated, such as the ITU-T G.711 st...

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): G10L19/08G10L25/30G10L25/45
CPCG10L19/08G10L25/30G10L25/45
Inventor 闵刚张长青解云虹谭薇周怀军吴广恩刘向阳
Owner NAT UNIV OF DEFENSE 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