Microphone array speech processing

Significant knowledge about microphone arrays has been gained from years of intense research and product development. There have been numerous applications suggested, for example, from large arrays (in the order of 100 elements) for use in auditoriums to small arrays with only 2 or 3 elements for hearing aids and mobile telephones. Apart from that, microphone array technology has been widely applied in speech recognition, surveillance, and warfare. Traditional techniques that have been used for microphone arrays include fixed spatial filters, such as, frequency invariant beamformers, optimal and adaptive beamformers. These array techniques assume either model knowledge or calibration signal knowledge as well as localization information for their design. Thus they usually combine some form of localisation and tracking with the beamforming. Today contemporary techniques using blind signal separation (BSS) and time frequency masking technique have attracted significant attention. Those techniques are less reliant on array model and localization, but more on the statistical properties of speech signals such as sparseness, non-Gaussianity, and non-stationarity. The main advantage that multiple microphones add from a theoretical perspective is the spatial diversity, which is an effective tool to combat interference, reverberation, and noise. The underpinning physical feature used is a difference in coherence in the target field (speech signal) versus the noise field. Viewing the processing in this way one can understand also the difficulty in enhancing highly reverberant speech given that we only can observe the received microphone signals.