Semi-Coprime Microphone Arrays for Estimating Direction of Arrival of Speech Sources

This paper evaluates the performance of semi-coprime microphone arrays (SCPMAs) for speech source direction of arrival (DOA) estimation based on the steered response power - phase transform (SRP-PHAT) algorithm. The SCPMA is an extension of the coprime microphone array (CPMA), which combines the outputs of three sub-arrays to reduce the impact of spatial aliasing and achieves performance comparable to that obtained from arrays using much larger numbers of microphones. The proposed approach considers two different processors to calculate the outputs from the sub-arrays and adapts the SRP-PHAT approach to these arrays. Simulations are conducted under anechoic and reverberant scenarios in a noisy room. Beam pattern and array gain results indicate that the SCPMA works better than the conventional CPMA at reducing the peak side lobe (PSL) level and total side lobe area while increasing the capability of amplifying the desired target signal and restraining noise from all other directions for typical frequencies of speeches. DOA Estimation results also show that the SCPMA achieves accurate DOA estimates in anechoic and low reverberant conditions, which is comparable to the equivalent full ULA, while the large side lobes in the beam pattern of the SCPMA lead to less accurate results in the highly reverberant environment.