Voice lie detection method based on convolution bidirectional short-time memory network

A long-short-term memory and speech technology, applied in speech analysis, instruments, etc., can solve problems such as algorithm does not have feature learning ability, information loss, and unclear mapping relationship, achieving good application prospects and improving performance

Active Publication Date: 2018-09-11
NANJING INST OF TECH
View PDF6 Cites 16 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0005] (1) The mapping relationship between many existing speech features and lies is not yet clear;
[0006] (2) The extraction process from original speech to speech features will inevitably lead to the loss of information, and it is unknown whether the lost information wil

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
  • Voice lie detection method based on convolution bidirectional short-time memory network
  • Voice lie detection method based on convolution bidirectional short-time memory network
  • Voice lie detection method based on convolution bidirectional short-time memory network

Examples

Experimental program
Comparison scheme
Effect test

Example Embodiment

[0048] The present invention will be further described below in conjunction with the accompanying drawings.

[0049] The speech polygraph method based on convolution two-way long-short-term memory network of the present invention comprises the following steps:

[0050] Step (A), performing unified normalization processing on the entire segment of speech;

[0051] The normalization of the data in this step is performed on the entire speech segment, not on each segment after cutting. The range after normalization is [-1, 1], and the speech value before and after normalization The physical meaning expressed by the zero place is unchanged, and it is all a silent segment, which is consistent with the meaning of zero padding of the unified data length when calculating the variable length data in the step (D);

[0052] Step (B), segmenting the voice of the unified normalization process according to the database label;

[0053] The database is a professional database established by ...

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 voice lie detection method based on a convolution bidirectional long and short time memory network. The method includes carrying out the unified normalization processing on the whole-section voice; segmenting the voice which is subjected to the unified normalization processing according to the database label; performing windowing and framing processing on the segmented voice; establishing a calculation mode of the variable-length data; introducing the convolution operation into the long and short time memory network; constructing a complete voice lie detection networkmodel; and training the voice lie detection network model, and performing lie detection and evaluation on the voice processed by windowing and framing. According to the invention, the convolution operation is introduced into the long and short time memory network, the complete voice lie detection network model is constructed, and deep learning is achieved, the features for lie detection are extracted from the original voice data to improve the lie detection performance, and the method has a good application prospect.

Description

technical field [0001] The invention relates to the technical field of speech lie detection, in particular to a speech lie detection method based on a convolutional bidirectional long-short-term memory network. Background technique [0002] We know that compared with the normal state, people will cause slight changes in sound pressure, tone, speech rate, pause time and vocal organs when lying, which will lead to changes in some characteristic parameters of speech. Therefore, by monitoring these changes, we can Enables lie analysis and detection. Although the research on lie detection has a long history, there are few relevant results focusing on the research on lie detection based on speech features as clues, so it has important theoretical research value. In addition, voice feature lie detection has the advantages of simple and concealed test process, remote detection of absent personnel, low equipment cost, etc., and has important application value. [0003] In 1991, Ekm...

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): G10L17/26G10L17/04G10L25/27G10L25/51
CPCG10L17/04G10L17/26G10L25/27G10L25/51
Inventor 谢跃梁瑞宇赵力包永强唐闺臣
Owner NANJING 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

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