Evaluation of multichannel speech signal separation with beamforming techniques

Recorded speech signal separation from noise and interference using multichannel microphone array is a prominent area of research in audio engineering. Beamforming technique calculates time delay among the array signals and projects a beam for maximum output power in the space. This paper presents two basic beamforming techniques to suppress the noise and interference from the recorded speech in noise signal for auditory training simulator design. MATLAB functions were used to simulate the recorded `.wav´ format speech in noise file by audiologist to build the Malay language based auditory training simulator training content. Results proved that Frost beamformer performs better than conventional time-delay beamformer in producing clearer speech signal to verify the word clarity for the speech verifying module in the auditory training simulator development.