The Research And Development Of Voice Conversion System Based On DSP

Digital signal processing technical is developing rapidly, and as a mainand important branch of it, voice signal processing contains voice analyze,voice enhancement,voice communication and so on. But super processorwas needed for the application of these technical. Traditional processors,forexample X86 series produced by Inter company is not suitable for densenesscalculation situation. So DSP(Digital Signal Processor) appears for thisapplication. It can do one add-mult in one machine circle,read or write oneblock of memory muti-times in one machine circle, and it has abundanceperipheries on chip.Voice conversion is a method which transforms the source speech toconverted speech with character information of target speaker. Formantfrequency, bandwidth and pitch are chosen as voice character parameterswhile formant synthesis method is used to create converted voice.Experiment is carried out on voice conversion of 5 vowels frominternational phonetic symbol and speech. Evaluation to converted results isconducted with MFCC auditory spectral method.After learning the relative knowledge about voice signal processing,voice conversion basic technical and the development in this domain,we didsome research and trying for the application based on DSP. After realizingthe voice conversion system on PC which was design based on MFC,andapproving the correct feasibility of functions being used, we migrate thesystem to the hardware system based on processor TMS320C5402successfully。The final result proved that DSP is faster than PC whileprocessing same amount of signal. It means that DSP processor can doperfect for voice digital processing system, and can prove the improvementfor it's application.The full text is divided into six chapters altogether:Chapter one:Having introduced the background and meaning of subjectresearch mainly and summarized the research current situations anddevelopment trends of the voice conversion technique.Chapter two:Recommend basic principle and necessary knowledge thatdevelop correlation instrument, Including the basic principle of voiceconversion,the principle of pre-processing about voice signal,theestablishments and applications of conversion function,relative theory ofvoice synthesis.Chapter three:Overall design of the hardware design for voiceconversion system,including design scheme,hardware structure,technicalindicator and the question should pay attention during developing thesystem.Chapter four:Software design,researching and realization of the voiceconversion system. Firstly, Design processing based on MFC in VC (onekind of software design platform) was introduced. The arithmetic whichbeing choose was introduced especially. Secondly, the introduction of voicesynthesis arithmetic still was showed. Finally, you can see therecommendation of the migration and development way we used.Chapter five:Analyzing some of the function,results and it's accuracyof the final machine.Chapter six:The full text is summarized, including the research workand achievement of this text in an all-round way, analyzing the place to alsoneed to perfect further for the design, putting forward the correspondingsuggestion for this work.