Sound source classi̇fi̇cati̇ons usi̇ng spi̇ki̇ng neural networks

In this master of science thesis, we have worked on determining sound source positions via processing data obtained from microphone arrays. We have focused on phase transformed steered response power algorithm which is one of the most powerful techniques on sound source localization. To listen sound sources separately, focusing is needed which requires the exact position knowledge of source localization when microphone arrays are used, For this purpose, to evaluate position information and localization, classification is incorporated using a spiking neural network. Results obtained indicate very a good performance. Determined positions are quite comparable with the known positions of five talkers, thus calculated positional errors are in a few centimeters range