Font Size: a A A

Arm-based Embedded Isolated Word Speech Recognition System With Realization

Posted on:2010-06-13Degree:MasterType:Thesis
Country:ChinaCandidate:Z Y JiangFull Text:PDF
GTID:2208360278469268Subject:Signal and Information Processing
Abstract/Summary:PDF Full Text Request
Speech recognition technology is actually to allow the computers to "understand" human language,then in accordance with its meaning to implement the corresponding order,in order to achieve the service of humanity. With in-depth study of speech recognition, its technology has two main aspects: In one direction is large vocabulary continuous speech recognition system, mainly used in computer dictation machines, as well as the telephone network or the Internet with a combination of voice information service system.These systems are at the implementation on the computer platform; Another important direction of development is the applications of small and portable audio products.These applications are the use of dedicated hardware systems.And with the advent of post-PC era, the latter one,which will be the a very hot topic of the development of speech recognition technology and embedded systems crossover study,will further promote the speech recognition technology to the intelligent direction.In this paper, its main study is the implementation on ARM embedded platform for speech recognition system.According to the characteristics of embedded systems and actual needs of the system platform,this paper is doing the appropriate custom tailoring of the target platform,and the improvement and optimization of the speech recognition algorithms.At the same time in order to enhance the interactive system,this paper increases the control interface,and provides good practices of human-computer interaction for the practical application.In this paper,first of all,it does studies of embedded systems and embedded operating systems,and selects embedded Linux operating system through the actual comparison;then,it discusses the speech recognition technology,and uses Mel Cepstrum parameter as an extraction algorithm and DTW as a pattern matching of speech recognition for this system,according to actual requirements.At the same time, it implements the procedures of the system algorithms and the application control,which are transplanted to the target board to test the system.Finally,the overall test of the system,through the experimental results show that the system designed to achieve the desired portable,intelligent and a very good exchange purposes.
Keywords/Search Tags:speech recognition technology, embedded system, QTE platform, extraction of characteristics parameters, pattern matching
PDF Full Text Request
Related items