Microphone array proccesing

Abstract

Microphone arrays consist of multiple microphones functioning as a single directional input device: essentially, an acoustic antenna. Using sound propagation principles, the principal sound sources in an environment can be spatially located. Distinguishing sounds based on the spatial location of their source is achieved by filtering and combining the individual microphone signals. The location of the principal sounds sources may be determined dynamically by analyzing peaks in the correlation function between different microphone channels. As an acquisition device for speech, microphone arrays have a distinct advantage as they place few constraints on the user, freeing them from the need to ‘wear’ a microphone (as in lapel, headset, or mobile devices) or be near to and speaking towards the microphone (as in web cams and lecturn microphones, for example). Unfortunately, arrays also some disadvantages compared to individual microphone, including increased hardware and processing requirements due to the multiple channels, and acoustic losses due to the distance from the speaker. Research into microphone array processing is thus required to minimize these issues, enabling the advantages of microphone arrays to be exploited in practical systems. The more accurate the system, better results can be obtained. It is therefore important to choose and place the sensors in the most effectively. This is demonstrated in this paper, implementing an algorithm to estimate the direction of arrival and position of the sound source, in the most possible efficient way.