Micro-array beam forming and testing for speech recognition in industrial settings

Abstract

Beamforming is a signal processing technique used to enhance quality of speech recordings, which is done by isolating the speaker’s voice from surrounding noise. This is present in a variety of applications such as wireless communication, and acoustics. It encompasses the control of signals to target them towards a specific point, hence improving the strength of the signal and decreases the interferences. This technique is made possible by merging signals from a microphone array with varying delays and amplitudes to make a more focuses signal on a fixed direction, while simultaneously reducing background noise and interferences from other directions. This study compares the performance of a microphone array to a shotgun microphone, via beamforming and testing for speech recognition. Both the microphone array and the shotgun microphone use beamforming to enhance quality of speech recording and are commonly used for speech recordings. In this study, we evaluate both the microphone array and the shotgun microphone in a noisy and reverberant environment. Various speeches were recorded and with the recorded speech data, the signal-to-noise (SNR) and word error rate (WER) were evaluated.