Frequency Response Analytical Calibration Based on Spectral Flatness for Microphone Arrays

Many audio processing tasks using microphone arrays suffer from the impact of frequency response mismatch among microphones. To mitigate this problem, a frequency response analytical calibration method based on spectral flatness for microphone arrays is proposed in this article. Specifically, the dynamic calibration model, including two geometric states, is first applied to arrange the uncalibrated microphones. Then, the calibration signals corresponding to two geometric states are synchronized by estimating the time difference of arrival (TDoA). Next, a filter design method is presented in the frequency domain, from which the coefficients of calibration filters can be derived as closed-form solutions. Furthermore, a noise reduction approach via the spectral subtraction technique is given to cope with the adverse effect of noises. Finally, the ripples introduced by calibration filters are further suppressed through a spectral flatness-based criterion. Simulation and real-world experimental results illustrate that the proposed method can effectively compensate both the gain and phase errors among microphones in noisy and reverberant environments.