Combined Multisensor-Based Angle Clipping Algorithm and Multichannel Noise Removal Method for Multichannel Sound Localization

This article proposes an efficient, robust, and highly accurate positioning algorithm called generalized cross correlation—threshold-based angle clipping (GCC-TBAC), which estimates the direction of a signal source based on the signals received by multiple sensors. The algorithm is applied to sound localization in this study, using a circular microphone array composed of four microphones. As sound propagates through space, the distances from the sound source to each microphone vary, resulting in slight differences in the time it takes for the signal to reach each microphone. This difference is known as the arrival time delay. In typical microphone setups, abnormal delays or acoustic interference between multiple channels can cause sudden angle anomalies or drift in the localization. This study proposes a TBAC algorithm to address the issue of multichannel sound interference. The algorithm utilizes the GCC phase transform (GCC-PHAT) to estimate the time delay of arrival and then classifies multiple sets of angles received by the microphone array based on the root mean square error (RMSE) to remove problematic signals, thereby improving accuracy. Additionally, increasing the number of microphones can improve accuracy. Experimental results demonstrate that the new algorithm, when combined with multiple microphones, effectively improves the accuracy of localization. Therefore, the proposed clipping algorithm and the research on clipping localization uncertainty reported in this article are applicable to various signal processing applications.