Looking for breakthrough ideas for innovation challenges? Try Patsnap Eureka!

Dialect species recognition method based on extended convolutional neural network

A convolutional neural network and recognition method technology, which is applied in the field of dialect species recognition based on dilated convolutional neural network, and can solve problems such as poor real-time performance.

Inactive Publication Date: 2020-06-05
BEIJING UNIV OF TECH
View PDF3 Cites 7 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, because RNN uses a loop mechanism, the real-time performance is not strong, so this has become a major disadvantage

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
  • Dialect species recognition method based on extended convolutional neural network
  • Dialect species recognition method based on extended convolutional neural network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0020] The specific embodiments of the present invention will be further described below in conjunction with the accompanying drawings.

[0021] Aiming at the deficiencies of the prior art, the present invention proposes a neural network based on dilated convolution, using the traditional acoustic feature Mel frequency cepstral coefficient MFCC and the Mel filter bank log-mel filterbank as the input of the neural network model, using the convolution The translation, scaling and rotation invariance of the product can improve the accuracy of dialect species recognition on the basis of fewer parameters trained in the network.

[0022] In the present invention, the neural network based on dilated convolution is applied to the identification of dialect species, forming a method for identifying dialect species based on the neural network of dilated convolution, including the selection of data sets, the preprocessing stage, and feature extraction. stage, model building stage, trainin...

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 dialect species recognition method based on an extended convolutional neural network. Dialects have rich cultural deposits as a unique national culture. If the Chinese dialect recognition is systematized, the dialects are classified and summarized to determine the species of the dialects. Three dialects are mainly selected, a corpus is constructed, audio data is digitizedand preprocessed, radiation of lips to voice signals is removed, high-frequency resolution is improved, and it is guaranteed that the voice signals have short-time stability after framing. The methodcomprises the following steps: extracting an acoustic feature Mel frequency cepstrum coefficient MFCC, log-mel filter of an audio, training by using a convolutional neural network CNN based on extended convolution, and fusing the acoustic feature MFCC, log-mel filter into a residouble network ResNet; and storing the optimal model, arranging classification labels, and storing mapping between labelclassifications and dialects. The judgment of a person is simulated in a deep learning mode to distinguish the types of dialects. The result shows that the method can improve the recognition accuracyto more than 90%, and can be used in the fields of dialect accent recognition and the like.

Description

technical field [0001] The present invention relates to a dialect species recognition method in the field of deep learning language recognition and speech signal processing, in particular, to a dialect species recognition method based on an expanded convolutional neural network. Background technique [0002] Based on different regional cultural forms, various local dialects have been formed, and the dialects in many places have the characteristics of "different sounds for ten miles". The role of social education such as regional cultural promotion and cultural characteristics inheritance of dialects should be carried forward. It is necessary to use technology to protect some dialects that are fading out of people's sight and become a fortune. [0003] Dialect recognition is analogous to language recognition, and its purpose is to automatically determine the language category to which a piece of speech belongs. As a front-end processing technology for related language appli...

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): G10L15/06G10L15/08G10L19/16G10L25/18G10L25/24G10L25/30G10L25/45
CPCG10L15/063G10L15/08G10L25/18G10L19/167G10L25/24G10L25/30G10L25/45
Inventor 李肖贺国平
Owner BEIJING UNIV OF TECH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Patsnap Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Patsnap Eureka Blog
Learn More
PatSnap group products

Browse by: Latest US Patents, China's latest patents, Technical Efficacy Thesaurus, Application Domain, Technology Topic, Popular Technical Reports.

© 2024 PatSnap. All rights reserved.Legal|Privacy policy|Modern Slavery Act Transparency Statement|Sitemap|About US| Contact US: help@patsnap.com