Real-time Multi Source Speech Enhancement based on Sound Source Separation using Microphone Array

This paper proposes a Sound Source Localization method based on source separation algorithm technique, which can be applied to application of reverberant environment where multiple acoustic sources are present. In this paper, we proposed the study and implementation of sound source localization and sound separation algorithm which uses three sources with eight microphones system is presented. The proposed system uses the signals which are acquired by a set of microphone arrays at different positions are considered as an input source. The recorded signals are applied into the source localization to obtain the Direction Of Arrival (DOA) of each source pair. The phase angle between the sources are estimated based on TDOA algorithm and mixture of source signals are separated based on the radial bins which is given into the system and the particular direction of signal is separated from the mixed signals. The Separated output noise level is tested with the input speech signal and it is found that noise level is comparatively reduced with the applied input signal.