Towards low cost acoustic cameras for the Internet of Things
* Presenting author
In the context of Internet of Thing (IoT) accurate and low-cost sound source localization methods are of great interest, alone or combined with video images. In this work, we present an implementation of the Steered-Response Power Phase Transform (SPR-PHAT) method employing a small circular array of 10 cm consisting of 8 MEMS microphones all commanded by a small single-board computer (Rasperry Pi). The localization is performed both in azimuth and elevation in one side of the array. A camera emplaced in the center of the array can be also employed for multimodal localization. The evolution of the localization accuracy as more microphone pairs autocorrelacions are added in the estimation of likelihood is compared. An optimal number of autocorrelation pairs in the sense of accuracy/computational cost tradeoff is obtained.