Principle And Implementation Of Video Conference System For Automatic Positioning

Theories of speaker tracking for smart-conferencing system have been studied in this thesis, which mainly includes theories of passive acoustical localization, theories of time delay estimation. In order to prove the validity of the theories, computer simulation and experiments have been carried out. Finally, a precept of passive acoustical localization system has been presented, which involves hardware and software designing.Firstly, the characteristics of sound wave and speech signal have been discussed. According to characteristics of speech signal, short-time processing method has been discussed, which bottoms the following research.Secondly, passive acoustical speaker tracking models have been established. These models consist of three elements linear microphone array model for small assembly rooms, four elements microphone array model and five elements microphone array model for big amphitheaters. The algorithm of speaker tracking has been derived. In addition, the primary factors that may influent tracking precision have analyzed.Thirdly, three algorithm of time delay estimation have been researched, which include Cross-spectrum method, Generalized Cross Correlation method and LMS adaptive filter method. Upon this, computer simulation has been done with three-microphone array model.Fourthly, basing on theory discussing, passive acoustical localization experiments have been carried out with speech signal, sine sweep signal, single sine signal and sine impulse signal. Coherence has been achieved between experimental results and theory results.Finally, a precept of passive acoustical localization system has been presented, which involves hardware and software designing.