Microphone array calibration for robust adaptive processing

Significant interest exists in the use of microphone arrays for enhancing audio and speech signals. Various combinations of fixed array, adaptive array, time-domain and frequency domain processing approaches are being considered. Any well-designed sensor array processing system needs (either explicitly or implicitly) accurate modeling or measurement of the sensor responses to allow robust and effective signal enhancement. The article explores the limits that microphone array calibration places on robust speech enhancement. Modelled microphone array data is used to examine the effect of variations in array parameters on processing algorithms.