In acoustic signal processing, Direction Of Arrival estimation plays an essential part in microphone-array technology as a preprocess in source localization or noise-robust speech recognition. In this thesis a sound source localization method for home network system was studied.
Usually, sound that happened in electronic appliance is consisted of pulsed continuous wave and noise. Therefore we must estimate correct position of sound source using this signals. However, the performances of the localizing method based on MCPSP function drops greatly in the case of CW signal. Therefore this thesis proposed efficient TDOA estimation method for pulsed CW signal. This thesis estimates the sound source localization using MCPSP function that is calculated on frequency domain using small number of sensors.
In the proposed method we composed of new segment including a edge of pulse. This segment was computed by short-time energy detection. With theoretical representation the performances of the proposed method were analyzed under various environments. From the several experimental results, we can confirm the performances of the proposed method.
This thesis organization is as following. SectionⅡ is described method of localization. In section Ⅲ, the proposed method is described, and experimental results are discussed in section Ⅳ, Conclusion and comments are given in section Ⅴ.