Speaker recognition method and system based on multi-source attention network

A technology of speaker recognition and attention, applied in the field of speaker recognition, can solve problems such as improvement of recognition accuracy

Active Publication Date: 2021-07-06
SHANDONG NORMAL UNIV
View PDF3 Cites 1 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0004] In order to solve the deficiencies of the prior art, the present invention proposes a speaker recognition method and system based on a multi-source attention network; it solves the problem of recognition accuracy ca...

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
  • Speaker recognition method and system based on multi-source attention network
  • Speaker recognition method and system based on multi-source attention network
  • Speaker recognition method and system based on multi-source attention network

Examples

Experimental program
Comparison scheme
Effect test

Embodiment 1

[0036] This embodiment provides a speaker recognition method based on a multi-source attention network;

[0037] Such as figure 1 As shown, the speaker recognition method based on multi-source attention network, including:

[0038] S101: Extract gender features of the speech segment to be recognized; extract accent features of the speech segment to be recognized;

[0039] S102: Extract the timbre features of the speech segment to be recognized based on the CNN network of the trained multi-source attention network;

[0040] S103: A gender attention network based on the trained multi-source attention network, using gender features and timbre features to construct gender auxiliary features;

[0041] S104: An accent attention network based on the trained multi-source attention network, using accent features and timbre features to construct accent auxiliary features;

[0042] S105: Perform speaker recognition by combining the timbre feature, gender auxiliary feature and accent a...

Embodiment 2

[0154] This embodiment provides a speaker recognition system based on a multi-source attention network;

[0155] Speaker recognition system based on multi-source attention network, including:

[0156] Gender and accent feature extraction module, which is configured to: extract the gender feature of the speech segment to be recognized; extract the accent feature of the speech segment to be recognized;

[0157] The timbre feature extraction module is configured to: extract the timbre features of the speech segment to be recognized based on the CNN network of the trained multi-source attention network;

[0158] A gender auxiliary feature construction module, which is configured to: use gender features and timbre features to construct gender auxiliary features based on the gender attention network of the trained multi-source attention network;

[0159] An accent auxiliary feature construction module, which is configured as: an accent attention network based on the trained multi-s...

Embodiment 3

[0165] This embodiment also provides an electronic device, including: one or more processors, one or more memories, and one or more computer programs; wherein, the processor is connected to the memory, and the one or more computer programs are programmed Stored in the memory, when the electronic device is running, the processor executes one or more computer programs stored in the memory, so that the electronic device executes the method described in Embodiment 1 above.

[0166]It should be understood that in this embodiment, the processor can be a central processing unit CPU, and the processor can also be other general-purpose processors, digital signal processors DSP, application specific integrated circuits ASIC, off-the-shelf programmable gate array FPGA or other programmable logic devices , discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor, or the processor may be any conventional processor, or...

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 speaker recognition method and system based on a multi-source attention network. The method comprises the steps of extracting gender features of a speech segment to be recognized; extracting accent features of the speech segment to be recognized; based on a CNN network of the trained multi-source attention network, extracting tone features of the speech segment to be recognized; on the basis of a gender attention network of the trained multi-source attention network, constructing gender auxiliary features by using the gender features and tone features; based on an accent attention network of the trained multi-source attention network, constructing accent auxiliary features by utilizing the accent features and tone features to ; and combining the timbre feature, the gender auxiliary feature and the accent auxiliary feature to identify the speaker. The speaker recognition method is designed by simulating the thinking mode of human for speaker recognition, so that the recognition effect is better, and particularly for speakers which are extremely easy to cause confusion, the speaker recognition method can better distinguish the speakers.

Description

technical field [0001] The invention relates to the technical field of speaker recognition, in particular to a speaker recognition method and system based on a multi-source attention network. Background technique [0002] The statements in this section merely mention the background technology related to the present invention and do not necessarily constitute the prior art. [0003] Speaker recognition refers to the identification of a speaker's identity based on a person's voice, which has received widespread attention in recent years. Speaker recognition has broad application prospects and can be widely used in security monitoring, speaker logs and other fields. In recent years, the deep neural network model combined with the self-attention mechanism has been increasingly used in the field of speaker recognition. This method maps a set of input features at the input end of the neural network to two sets of intermediate features through the network, using two Group interme...

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): G10L17/04G10L17/18G10L25/24G06N3/04G06N3/08
CPCG10L17/04G10L17/18G10L25/24G06N3/08G06N3/045
Inventor 冷严李文静王荣燕唐勇孙建德齐广慧李登旺万洪林
Owner SHANDONG 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