Style migration method of music with human voice

A music and style technology, applied in the field of data processing, can solve problems such as unsatisfactory effects and the inability to achieve the number of cover songs by singers

Inactive Publication Date: 2020-10-23
JINLING INST OF TECH
View PDF0 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the effect of these algorithms on popular music with human voices is not satisfactory.
At present, a large number of songs are covered in various styles, but the number of covers by singers is fa

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
  • Style migration method of music with human voice
  • Style migration method of music with human voice
  • Style migration method of music with human voice

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0056] Below in conjunction with accompanying drawing and specific embodiment the present invention is described in further detail:

[0057]The invention provides a music style migration method with human voice, based on CycleGAN and WaveNet decoder, to solve the current music style conversion field. Most image style transfer algorithms perform poorly on music style transfer. The algorithm performed even worse on style transfer in music with vocals.

[0058] The overall model architecture diagram of the present invention is as follows figure 1 As shown, the audio image with human voice generated by deconvolution is as follows figure 2 As shown, the nearest neighbor interpolation is used to generate the audio image with human voice as shown in image 3 As shown, the algorithm of the patent invention has been tested on the computer programming environment, and we have verified the correctness and feasibility of our algorithm through the experiment. The specific configuratio...

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 style conversion method of music with human voice. According to the method, CQT features and Mel spectrum features of music are extracted, then style migration is carried outon pictures mapped by the CQT features and the Mel spectrum features by adopting a CycleGAN, and finally the migrated spectrograms are decoded through a WaveNet decoder to form the music. And finally, the audio quality and the style migration rate after the style migration of the music with the human voice are realized and improved. The invention provides a style migration method of music with human voice, which is based on a CycleGAN and a WaveNet decoder and solves the problems that most image style conversion algorithms show bad performance in music style conversion and the algorithm performance is worse in style conversion of music with human voice in the current music style conversion field.

Description

technical field [0001] The invention belongs to the technical field of data processing, and in particular relates to a style transfer method of music with human voice. Background technique [0002] Over the past three years, neural style transfer has continued to grow into a thriving field of research. Increasing activity within this field of research is driven by scientific challenges and industrial needs. Style transfer has broad application prospects including social, assisted user creation and entertainment applications. [0003] Musical style transfer is another field attempt of style transfer algorithms. Since music is a time-based segment and has many components, the extraction of features is more complicated, and the connection between features is more complex and tight. At present, most of the algorithms for images in the academic circle are directly applied to the music style transfer based on the image style transfer algorithm, and most of the music is pure mus...

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): G10L25/18G10L25/24G10L25/30G10L25/48G06F16/65G06N3/02G06N3/08
CPCG06N3/02G06N3/08G10H2210/031G10H2210/041G10H2250/131G10L25/18G10L25/24G10L25/30G10L25/48G06F16/65
Inventor 叶洪良朱皖宁
Owner JINLING 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