Font Size: a A A

Hmm-based Voice Recognition System

Posted on:2006-08-25Degree:MasterType:Thesis
Country:ChinaCandidate:L TangFull Text:PDF
GTID:2208360185963637Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
Speech Recognition is trying to make machines understood nature language and execute relevant task. Speech Recognition technology is concerned with many research fields of varied subjects. It has close relationship with acoustics,phonetics,linguistics,artificial,intelligence,digital signal processing,pattern recognition as well as optimization theory etc. All this research fields have contributed a lot to the development of speech recognition. Some problems due to the essential characteristic of speech, which include levity,dynamic,instantaneous,continuum and so on, cause difficulties in recognition practice. The process of computer speech recognition is almost the same as that of human's. Now the mainstream of speech recognition technology is based on the basic theory of statistic mode recognition. Considering the practical application, how to improve the performance of speech recognition system is a important issue in speech recognition technology.Our purpose of research is to realize an isolated word command recognition system based HMM . The author's work can be summarized as follows:1. Accomplished an entire program which include speech input,pre-process,training,recognition and result output.2. In this paper, there is brief introduction on the theory of Hidden Markov Model(HMM). In order to put HMM into practical speech recognition applications, three important problems of HMM have to be solved.3. Then comes the whole structure and frame of our isolated word recognition system, including speech acoustic analyzing,feature extraction,acoustic modeling and recognition strategy in HMM, at the end of this chapter we get the result of isolated word recognition system..4. Considering the practical recognition system application, how to acquire satisfactory performance under noisy environment is an important part of this article. We present a robust speech feature extraction method based on properties of the human auditory system and speech enhancement algorithm. The experimental results show that the performance of our system can be improved greatly by the proposed method under noisy environment.
Keywords/Search Tags:Speech recognition, Hidden Markov Model(HMM), Noise, Masking properties
PDF Full Text Request
Related items