Speech Recognition System Of Speaker-independent And Isolated Words Based On DSP

Speech recognition technology involves phonetics, artificial intelligence, digitalsignal processing technology, physiology, mathematical statistics, and otherinterdisciplinary knowledge and it is the most convenient way of man-machinecommunication. Recently, it is a major research focus that applying the speechrecognition to mobile devices, smart home and other areas. The problem we areaddressing in this paper is that of recognizing isolated word speech on DSP.Firstly, we create the template library of speech automatically based on theCambridge HTK tool sets which is mainly used for speech recognition. The templatelibrary has a simple training process and is free of expansion. Besides, the recognitionrate has improved as we can approach the optimal HMM parameters.Then we study sampling, digitization, pre-emphasis processing algorithm,feature extraction algorithm, endpoint detection algorithm, speech recognitionalgorithm of the voice signal. To cater for the recognition rate will decline as wecannot identify the beginning and ending points of the speech signal very well in anoisy environment, an endpoint detection algorithm combined by automaticallyadjusting thresholds algorithm and enhancing the zero-crossing rate algorithm isutilized, and it is shown to perform with good precision on finding the start and endpoints of voice signal, which is simulated by MATLAB. To enable high computingspeed and good real-time to be performed, we use the improved Mel cepstralcoefficients for characteristic parameters. Finally we have realized the HMM-basednon-specific speech recognition system successfully.Lastly, we realize the non-specific isolated word speech recognition systembased on CCS which is the software development tool of DSP andICETEK-DM642-PCI platform which consists of TMS320DM642processor core andTLV320AIC23audio codec.The experiment results show that the recognition rate is up to95%in this system.