An impact and noise resistance process of limiting observation probability minimum value in a speech recognition system

A technology of observation probability and speech recognition, applied in speech recognition, speech analysis, instruments, etc., can solve the problem of not considering front-end data reconstruction and recognizer matching, limiting the practicability of speech recognition system, affecting recognition speed, etc.

Inactive Publication Date: 2003-12-31
TSINGHUA UNIV
View PDF0 Cites 8 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

[0043] (1) Additional operations brought about by impact noise detection and data reconstruction will seriously affect the recognition speed
[0044] (2) The front-end data reconstruction and the matching of the recognizer are not considered, which limits the improvement space of the recognizer performance
[0045] Therefore, such methods still limit the practicality of speech recognition systems

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
  • An impact and noise resistance process of limiting observation probability minimum value in a speech recognition system
  • An impact and noise resistance process of limiting observation probability minimum value in a speech recognition system
  • An impact and noise resistance process of limiting observation probability minimum value in a speech recognition system

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0123] Such as Figure 5 As shown, at the beginning of the main program, the parameters of the HMM are first read in, and these parameters are obtained through training. According to these parameters, the dispersion index of each dimensional feature can be calculated, so as to obtain the sensitivity of each dimensional feature to noise, and thus divide the features according to the sensitivity, and then calculate the minimum value limit of each part of the feature when calculating the probability threshold. The voice file to be recognized is shown in the file list. The program reads the voice data to be recognized according to the list, and then performs feature extraction. In the actual system built, the extracted features are 13-dimensional MFCC and 13-dimensional ΔMFCC. The search for the optimal state sequence is an iterative algorithm based on Viterbi decoding. When calculating the observation probability of each frame of speech features, the minimum probability value of...

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

An impact and noise resistance process of limiting observation probability minimum value in a speech recognition system characterized by that, in the optimum state sequence searching stage of the implicit Malkov model probability statistics recognition method, first using the scattering index to classify the noise sensibility degree of each dimension of the voice characteristics, then using threshold to perform minimum value limitation to the observation probability of the sensible characteristics, so as to diminish effectively the influence caused by noise abatement, and at the same time keep more information useful for recognition, which can substantially improve the function of the speech recognition system in impact and noise.

Description

technical field [0001] The anti-shock noise method of limiting the lowest value of observation probability in the speech recognition system belongs to the field of speech recognition technology, especially the field of probability and statistics recognition method of hidden Markov model (Hidden Markov Model, namely HMM). Background technique [0002] Speech recognition is a technology in which a machine converts speech signals into corresponding text words or commands through the process of recognition and understanding. With the rapid development of information technology, people are getting more and more information. At the same time, they also hope to have a friendly man-machine interface to realize easy communication with machines. Natural language is the most flexible, effective, and convenient communication medium for human beings, so the realization of communication with machines through speech recognition has naturally become the goal pursued by people. Speech recog...

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): G10L15/14G10L15/20
Inventor 丁沛曹志刚
Owner TSINGHUA 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