Implementation of Real-Time Sound Source Localization using TMS320C6713 Board with Interaural Time Difference Method

Abstract

Innovations in robotics have been widely used and developed to facilitate various lines of human life. Robots are often designed to have human-like abilities (such as by implanting hearing abilities). However, many challenges need to be overcome in the application process. One of them is related to sound localization (determination of the sound source direction). Research related to sound localization has indeed been carried out by many researchers before. However, previous research still tends to use a lot of array microphones for the sound localization process, and the delay in sound detection still needs to be considered. Therefore, we are trying to build a real-time sound source localization system using the Interaural Time Difference Method. In this study, two array microphones were used for sound source detection, and a DSP board (TMS320C6713) was used for data processing so that this system could detect sound sources quickly and accurately. Based on the test results, this system can detect the source and direction of sound in real-time. It can be seen from the value of the detection delay, which ranges from 3. 05-27.08ms. In addition, the system can also predict sound direction quite well even though at measurement angles of more than -80° and 80°, there is still an estimation error (around 3.61% - 4.76%). Meanwhile, at angles of measurement less than -80° and 80°, the accuracy of the estimation of the direction of sound reaches 100%.