The Design And Realization Of Speech Detection System Based On DSP

With the rapid development of digital hardware technology, DSP, in the technology of digital signal processing had made the theory and methods of digital signal processing come true in practical application. Speech detection was a fundamental and critical issue in speech signal processing. Effective speech detection can not only reduce the processing time and improve the processing speed but also avoid the interference of noise from silence part in speech, so the following recognition would enhanced greatly. Therefore the research of speech detection had great important practical significance. This paper discussed the design and realization of speech dectection based on DSP.The dissertation firstly intrduced hardware electrocircuit design of the speech detection system, the function of A/D and D/A was implemented by TLV320AIC23, the acquisition and disposal of data was realized by TMS320VC5416, while PCI2040 realized the PCI local bus protocol. Secondly, relative theory and exploitation of Driver was intrduced. Lastly, the dissertation discussed the research for several algorithms of speech detection and its realization in the system.System software consisted of three parts: DSP procedures, Driver and application procedures. DSP procedures mainly take charge of speech signal detection and commucate with application through interrupt mode. Driver was an interface which connected DSP to host. The main effect of application was data acqusition from the memory of DSP and storeed the data in the disk.Experiment was composed of two parts: in the experiment of real-time speech acquisition, the result showed that no data was lost and there was a smaller distortion; in the experiment of speech detection, we could obtain satisfactory results when the SNR is high.