Font Size: a A A

Research And Implementation Of The Speech Recognition Technology Based On DSP

Posted on:2010-11-28Degree:MasterType:Thesis
Country:ChinaCandidate:T LvFull Text:PDF
GTID:2178360278462269Subject:Traffic Information Engineering & Control
Abstract/Summary:PDF Full Text Request
In recent years, speech recognition is a relatively active area of research, has been a lot of applications in embedded systems, making the original manual work needed to use voice can be easily completed. The use of voice as a means of human-computer interaction for users is a very convenient, very natural way, at the same time the small device also omitted in order to reduce the volume.In this paper, to put forward a improved speech recognition algorithm based on the high-speed DSP platform, for China's most speech recognition products of low accuracye, real-time poor, as well as the noise of the weak, and other defects, by analyzing the speech recognition of the dynamic and the level of research at home and abroad.In this paper, the realization of speech recognition by the major hardware and software components algorithm. TI's high-speed digital signal processor chip TMS320VC5416 used as a system platform, through a microphone to realize the voice signal acquisition, and then by the A / D conversion chip to receive and send signals to the acquisition of the digital signal processor memory, in accordance with the complex programmable logic device the output timing for processing, the final results of the identification. Paper focuses on improving the algorithms and debugging: The main voice signal from the acquisition, preprocessing algorithm, Mel Cepstrum extraction, such as the composition of the two-stage endpoint detection. First prepared in the MATLAB algorithm simulation environment, and then use the algorithm C / C + + and assembly language into the preparation of DSP software development platform CCS debugging, the final programming to the DSP, and use VC + + to prepare for man-machine interface DSP with serial communications. And made use of to improve the algorithm so that the system's performance in a timely manner has been a great improvement, and the recognition rate has been a certain improvement in a noisy environmental.Finally, the paper summarized the research topic, according to experience and gives the System Design in the direction of several possible improvements.
Keywords/Search Tags:Speech Recognition, Endpoint detection, DSP, Mel Cepstrum parameters
PDF Full Text Request
Related items