Unlock instant, AI-driven research and patent intelligence for your innovation.

A Method of Speech Recognition Using SVM Optimized by Variation Fish Swarm Algorithm

A fish swarm algorithm and speech recognition technology, applied in speech recognition, speech analysis, instruments, etc., can solve the problems of the support vector machine parameters affecting the classification performance, the recognition rate is not significantly improved, and there is no theoretical method, so as to reduce the structure and calculation. Complexity, speed of convergence, effect of enhanced coordination

Inactive Publication Date: 2019-08-09
TAIYUAN UNIV OF TECH
View PDF5 Cites 0 Cited by
  • Summary
  • Abstract
  • Description
  • Claims
  • Application Information

AI Technical Summary

Problems solved by technology

However, the problem at this stage is that the setting of support vector machine parameters will seriously affect its classification performance, and so far there is no more effective theoretical method for guidance
However, this method does not significantly improve the recognition rate under certain conditions of the speech library, and the optimization time for the overall best parameters is still long

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 Method of Speech Recognition Using SVM Optimized by Variation Fish Swarm Algorithm
  • A Method of Speech Recognition Using SVM Optimized by Variation Fish Swarm Algorithm
  • A Method of Speech Recognition Using SVM Optimized by Variation Fish Swarm Algorithm

Examples

Experimental program
Comparison scheme
Effect test

Embodiment Construction

[0037] The present invention uses the windows 7 system as the program development software environment, uses MATLABR2010b as the program development platform, collects voice samples through the microphone as input data, and then the voice recognition system uses a series of algorithms to carry out preprocessing and feature extraction successively on the input voice signals (here The process adopts the existing technology), and the feature extraction of speech recognition is to extract the parameter set that can represent the speech signal from the speech signal, assuming that these parameters are the 10-dimensional feature matrix obtained from each frame of speech signal, as shown in Table 1 ( In this example, the feature matrix of 10 isolated words pronounced by three people under the condition of signal-to-noise ratio of 0db is used as the training set sample input), and finally the feature matrix train_data and the category label train_label corresponding to the feature matri...

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 the technical field of speech recognition, provides a method of speech recognition using an SVM optimized by a mutated fish swarm algorithm. The visual field of mutated fish is improved adaptively. In the initial stage of iteration of the mutated fish swarm algorithm, the gen value is smaller, the visual field is larger, and individual fish can be updated within the entire traversal range. In the later stage of iteration, as the gen value increases, the visual field gradually narrows, and individual fish is optimized and updated within a small range, which is more conductive to finding a global optimal value accurately. By reducing the use of parameters and simplifying the behavior mode in the artificial fish swarm algorithm, the structure and the computational complexity of the mutated fish swarm algorithm are reduced, the problem that the algorithm gets trapped into local optimum in the process of parameter optimization is avoided effectively, and convergence is accelerated. When the algorithm is applied to a speech recognition system, the speech recognition rate is higher, and convergence is faster.

Description

technical field [0001] The invention relates to the technical field of speech recognition. Background technique [0002] The voice is converted into an electrical signal through the microphone and then added to the input of the recognition system. After preprocessing, the features of the speech signal are extracted. First, the required template is established on this basis. The process of establishing the template is called the training process. The next process of matching the newly extracted features with templates is called the recognition process. That is, according to the overall model of speech recognition, the characteristics of the input speech signal are compared with the existing speech templates, and a series of optimal templates matching the input speech are found according to a certain search and matching strategy. Then, according to the definition of the template number, the recognition result of the computer can be given by looking up the table. [0003] Be...

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): G10L15/08G10L15/10G10L15/02
CPCG10L15/02G10L15/08G10L15/10
Inventor 白静朱文静薛珮芸张雪英
Owner TAIYUAN UNIV OF TECH