View article

This chapter proposes a direction of arrival (DOA) estimation of an acoustic source based on support vector machine (SVM) with uniform linear array (ULA). The SVM has been trained using correlation features of the microphone’s signals of the ULA. The DOA estimation accuracy of SVM model has been compared against the standard algorithms, i.e. delay and sum (DAS) beamformer as well as recurrent neural network (RNN) model and from that it has been observed that SVM-based DOA estimation outperforms DAS beamformer in all cases and has better performance than RNN model for low values of signal-to-noise ratio (SNR).