A Short Speech Speaker Recognition Method Based on Sparse Representation

A technology of speaker recognition and sparse representation, which is applied in the field of short speech speaker recognition based on sparse representation, can solve the problems of semantic information mismatch, speaker model can not effectively improve the accuracy of recognition, etc.

Inactive Publication Date: 2016-05-11
UNIV OF ELECTRONICS SCI & TECH OF CHINA
View PDF4 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0006] Aiming at the prior art, the technical problem mainly solved by the present invention is to provide a short speech speaker recognition method based on sparse representation, which can not effectively improve the existing technology when the semantic information does not match and the speaker model does not match. The problem of recognition accuracy

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
  • A Short Speech Speaker Recognition Method Based on Sparse Representation
  • A Short Speech Speaker Recognition Method Based on Sparse Representation
  • A Short Speech Speaker Recognition Method Based on Sparse Representation

Examples

Experimental program
Comparison scheme
Effect test

Embodiment

[0052] Such as figure 1 As shown, a short speech speaker recognition method based on sparse representation, including the following steps:

[0053] Step 1: Preprocessing all speech samples, mainly including pre-emphasis, frame-based windowing, endpoint detection, and then extracting MFCC and its first-order difference coefficients as features;

[0054] Step 2: Train the Gaussian background model from the background speech library, and extract the Gaussian super vector as the secondary feature;

[0055] Step 3: arrange the Gaussian supervectors of the training speech samples together to form a dictionary;

[0056] Step 4: Use the sparse solution algorithm to solve the representation coefficients, reconstruct the signal, and determine the recognition result based on the minimized residual.

[0057] in such as figure 2 As shown, the first step includes steps S11, S12, S13 and S14, specifically as described below:

[0058] S11: Pre-emphasis, the high-frequency speech signal ...

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 short speech speaker recognition method based on sparse representation, belongs to the technical field of speech signal processing and pattern recognition, and aims to solve the problem of low recognition rate of the existing method under the condition of limited speech data. It mainly includes the following steps: ①Preprocess all speech samples, and then extract Mel cepstral coefficients and their first-order difference coefficients as features; ②Train the Gaussian background model from the background speech library, and extract Gaussian supervectors as secondary features; ③Arrange the Gaussian supervectors of the training speech samples together to form a dictionary; ⑤Use the sparse solution algorithm to solve the representation coefficients, reconstruct the signal, and determine the recognition result based on the minimized residual error. The Gaussian supervector obtained by self-adaptation in the present invention can greatly alleviate the problem of insufficient performance of the speaker's personality characteristics caused by limited speech data; the reconstruction residual of sparse representation can be used for classification, and the speaker can be dealt with due to the mismatch of semantic information. The problem of model mismatch.

Description

technical field [0001] The invention belongs to the technical field of speech signal processing and pattern recognition, in particular to speaker recognition technology under short speech conditions, and in particular to a short speech speaker recognition method based on sparse representation. Background technique [0002] Speaker recognition technology refers to using the voice features of speakers to identify their identities. It belongs to the category of biometric authentication technology and is widely used in judicial identification, Internet security, and military and national defense fields. There are still many problems in the practical process of speaker recognition technology, among which the training recognition problem under short speech conditions has attracted widespread attention. [0003] At present, the Gaussian Mixture Model-Universal Background Model (Gaussian Mixture Model-Universal Background Model, GMM-UBM) is generally used for short speech problems a...

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 Patents(China)
IPC IPC(8): G10L17/00G10L17/04
Inventor 程建黎兰苏靖峰周圣云李鸿升
Owner UNIV OF ELECTRONICS SCI & TECH OF CHINA
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