Frost Beamformer is used for speech enhancement applications. The beamformer implementation readily adapts to variations in room acoustic reverberation time and is immune to reverberation. The beamformer software package includes a beamforming algorithm to dynamically steer a beam at the source location. A source localization algorithm is also available to provide the beamformer with the source location coordinates.
VOCAL’s Frost beamformer software may be licensed as a standalone algorithm or library with custom solutions available. Contact us to discuss your speech application requirements.
Frost Beamformer Software
The Frost Beamformer[1], as illustrated in the figure, is an adaptive beamformer with a structure that is very similar to a delay and sum beamformer. Each microphone is filtered by a FIR filter with J weights. The beamformer output y(k) is formed by a summation of all the FIR filter outputs. This beamformer employs a linearly constrained minimum variance (LCMV) filter.
The Frost Beamformer performs well in speech enhancement applications where the beamformer output SNR is observed to have small changes with variations in room acoustic reverberation time [2]. This shows that this beamformer is immune to reverberation. This adaptive beamformer also provides null steering on interferences for improved spatial speech discrimination.
More Information
References
[1] O.L. Frost III, ”An algorithm for linearly constrained adaptive array processing,” Proc. IEEE,vol. 60, pp 926-935, August 1972.
[2] J. Benesty, J. Chen, and Y. Huang, ”Microphone Array Signal Processing”, Berlin, Germany, Springer-Verlag, 2008.