Demo: Sound Localization using Smartphone

Abstract

Smartphones based sound direction estimation can be helpful in many situations. For example, a deaf person in a meeting room can look at the smartphone to find out which direction the speaker is in and then he can look in appropriate direction to read lips/gestures of the speaker. Many smartphones today come with two built-in microphones located at physically different positions. This difference in position can cause time difference of arrival (TDOA) of sound on both microphones. Value of TDOA for two microphones may vary depending on the location of sound source with respect to the smartphone. This time difference of arrival can be used to estimate incoming sound direction with respect to smartphone. Challenges involved in angle estimation arise mainly because of heterogeneous characteristics of different types of sounds, small distance between two microphones on the smartphone and different positions of microphones on different devices. In this work we implemented TDOA based angle estimation for white noise as sound source. We look at this work as a first step towards achieving application described earlier. TDOA based techniques have been used before for angle estimation as described by Murray et.al in [?]. Their work however requires dedicated hardware and hence need some preparatory setup. More than two microphones have also been used for angle estimation as described in [?]. This technique involved using 6 microphones placed at different heights. Our approach uses a commodity smartphone and requires no preparatory setup. We don’t use any signal processing and there are no requirements for internet connectivity. Application computes TDOA using signal cross-correlation and then maps the TDOA to appropriate angle. Angle is measure anti-clockwise with respect to the camcorder of the