Removal of Noise from Speech - A case study of a construction machine

Abstract: This report proposes a solution to allow speech to enter a construction machine cabin while removing noise sources. Examples of noises sources include, engine noise, traffic or other construction activities. Survey of the possible methods to remove noise from speech was conducted, however, beamforming was the technique chosen. Beamforming uses multiple microphones to spatially filter a sound source and increase the signal to noise of the incoming signal. The method of how microphone locations were chosen are discussed by use of an analytical model. Additionally in this report, a real-time signal processing system and its components are described. Beamforming proved to increase the signal to noise ratio compared to a single microphone, however, there was an unintended amplification of the noise and introduction of artefacts that decreased the listening quality.