Optimized Filterbank Frost Beamforming

J. Hong, J. Park, S. Jeong, S. Lee, and M. Hahn (Korea)


Speech enhancement, microphone array, filterbank, Frost beamforming


Beamforming is intelligent spatial filtering which extracts only target signals from noisy environment using microphone arrays. Fixed beamforming is simple and easy concept but does not show good performance in real noisy situations. As an adaptive method, Frost (man’s name) beamforming is a good alternative. The error correction of Frost beamforming is a very effective feature in the aspect of performance. Furthermore, to adjust elements of each filterbank such as filter length and learning rate, Quadrature Mirror Filter (QMF) is implemented. As QMF is attached to Frost beamforming, the filter lengths and learning rates of the beamformer are controlled in each band. The meaning of optimization is the adjusting of filter lengths and learning rate to the best condition in the aspects of performance. The performance is measured by SNR (Signal to Noise Ratio) gain and Bark scale Spectral Distortion (BSD).

Important Links:

Go Back