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

Discriminative feature extraction method applied to language identification

A technology of feature extraction and language recognition, which is applied in speech recognition, speech analysis, instruments, etc., and can solve problems such as different language discrimination of different phonemes

Active Publication Date: 2017-01-04
NAT COMP NETWORK & INFORMATION SECURITY MANAGEMENT CENT +1
View PDF5 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to solve the deficiency that different phonemes have different language discriminations in existing features based on phoneme posterior probability, thereby providing

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
  • Discriminative feature extraction method applied to language identification
  • Discriminative feature extraction method applied to language identification
  • Discriminative feature extraction method applied to language identification

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0056] The present invention will be further described now in conjunction with accompanying drawing.

[0057] The discriminative feature extraction method of the present invention includes two stages, one is a training stage and the other is a testing stage. The work to be done in the training phase is to use the speech data in the training set to calculate the F-ratio index (F-Ratio) and use the speech data in the training set to train the PCA (Principal Component Analysis) matrix. The work to be done in the test phase is: use the F-Ratio index and PCA matrix obtained in the training phase to extract the features of the speech to be tested, and the extracted features are language discriminative.

[0058] The work to be completed in the training phase and the testing phase will be described respectively below.

[0059] 1. Training stage

[0060] Assuming that there are data of M languages ​​in a training set, and each language has N sentences (the number of sentences in each...

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 relates to a discriminative feature extraction method applied to language identification, which comprises the steps of calculating an F-ratio index for training a frame-level phoneme posterior probability feature vector of centralized voice data in a training stage, wherein the F-ratio index reflects the contribution of each dimension in the phoneme posterior probability feature vector for language discrimination; and extracting a phoneme posterior probability feature vector for voice to be tested in a testing stage, and performing feature weighting on the extracted phoneme posterior probability feature vector according to the F-ratio index.

Description

technical field [0001] The invention relates to the field of computer language recognition, in particular to a discriminative feature extraction method applied to language recognition. Background technique [0002] Language recognition refers to the technology that a computer automatically determines or confirms the language type of a speech. This is a technology that enables large-scale cross-lingual speech recognition applications, such as spoken language translation, spoken document retrieval, and more. It is also a research hotspot in information extraction in the field of intelligence and security. The key to language recognition technology is to use scientific methods to measure the individuality of different languages. Cognitive experiments have shown that differences between languages ​​can be reflected by features at different levels, including acoustic, phoneme, prosodic, morphological, and syntactic features. [0003] Acoustic layer features are usually extract...

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): G10L15/02
Inventor 袁庆升周若华云晓春张健陈训逊颜永红徐杰李锐光
Owner NAT COMP NETWORK & INFORMATION SECURITY MANAGEMENT CENT
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