Recursive jump connection deep learning music automatic generation method based on layer standardization

An automatic generation and skip connection technology, applied in the computer field, can solve the problem that Transformer does not have recursive characteristics

Pending Publication Date: 2021-11-26
SHAANXI NORMAL UNIV
View PDF0 Cites 4 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, since music is a work of art with contextual characteristics, Transformer does not have recursive characteristics

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
  • Recursive jump connection deep learning music automatic generation method based on layer standardization
  • Recursive jump connection deep learning music automatic generation method based on layer standardization
  • Recursive jump connection deep learning music automatic generation method based on layer standardization

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0057] Taking 200 Japanese pop music piano pieces as an example, as a training set, the recursive skip connection deep learning music automatic generation method based on layer standardization of the present embodiment is composed of the following steps (see figure 1 ).

[0058] (1) Collect instrument digital interface data

[0059] Collect music files of keyboard instrument and musical instrument digital interface type with fixed music style as the training of automatic music generation network.

[0060] (2) Preprocessing the training set

[0061] The electronic score in the musical instrument digital interface file is represented by an event, and the event is divided into: pitch, sound intensity, sound duration, position, bar, rhythm, chord event, and the pitch event represents the beginning of the pitch of the musical instrument digital interface type music file , the intensity event represents the dynamic level of the note event - corresponding to the perceived loudness,...

Embodiment 2

[0097]Taking 200 selected Japanese pop music piano pieces as an example as a training set, the method for automatically generating music based on layer standardization-based recursive skip connection deep learning in this embodiment consists of the following steps.

[0098] (1) Collect instrument digital interface data

[0099] This step is the same as in Example 1.

[0100] (2) Preprocessing the training set

[0101] This step is the same as in Example 1.

[0102] (3) Build music automatic generation network

[0103] The music automatic generation network model consists of an input nesting layer 1, a position encoding layer 2, a multi-head attention layer 3, a first normalized summation layer 4, a multi-expert layer 5, a second normalized summation layer 6, a linear regression model 7, a logic The regression model is composed of 8 connections. The output of the input nesting layer 1 is connected to the input of the position encoding layer 2, the output of the position enc...

Embodiment 3

[0134] Taking 200 selected Japanese pop music piano pieces as an example as a training set, the method for automatically generating music based on layer standardization-based recursive skip connection deep learning in this embodiment consists of the following steps.

[0135] (1) Collect instrument digital interface data

[0136] This step is the same as in Example 1.

[0137] (2) Preprocessing the training set

[0138] This step is the same as in Example 1.

[0139] (3) Build music automatic generation network

[0140] The music automatic generation network model consists of an input nesting layer 1, a position encoding layer 2, a multi-head attention layer 3, a first normalized summation layer 4, a multi-expert layer 5, a second normalized summation layer 6, a linear regression model 7, a logic The regression model is composed of 8 connections. The output of the input nesting layer 1 is connected to the input of the position encoding layer 2, the output of the position en...

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 recursive jump connection deep learning music automatic generation method based on layer standardization. The method comprises the steps of collecting musical instrument digital interface data, preprocessing a training set, constructing a music automatic generation network, training the music automatic generation network and automatically generating a music file. According to the method, on the basis of the structure of the Transformer-XL neural network, the layer-standardized recurrent neural network and the multi-expert layer are introduced, so that the overall neural network performance is optimized, the condition of recursion disappearance or explosion is relieved, the learning ability of the neural network is enhanced, and the quality of the generated music is higher and is closer to the type of a training set. The method can be applied to the technical field of automatic music generation.

Description

technical field [0001] The invention belongs to the technical field of computers, and in particular relates to a music automatic generation method based on layer standardization-based recursive skip connection deep learning. [0002] technical background [0003] Music creation refers to the complex mental and skill production process for music professionals or composers to create musical pieces with musical beauty. The main way is to combine different syllables according to the time sequence relationship, such as melody and harmony, and organize them with appropriate rhythms to produce dynamic sound waves with special timbre and texture. Music creation is usually composed of composers who have received professional music training and education to create musical pieces with musical beauty, which is an extremely complex technical solution. [0004] With the widespread application of artificial intelligence deep learning algorithms in image recognition, video detection, natura...

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): G10H1/055G06N3/08
CPCG10H1/0553G06N3/08
Inventor 张玉梅李琦杨红红吕小姣
Owner SHAANXI NORMAL UNIV
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