(The fast and robust acoustic beamformer)
(Includes a few bugfixes found in 2.0 since it was released.)
(This version is still under development, some people have experienced problems in some files by using it)
Stable version of BeamformIt version 3. It is a MUCH cleaner implementation in C++ of the classical BeamformIt.
Some minor bugfixes and requested enhancements.
- BeamformIt-3.5.tgz (released June 9th, 2014)
This version includes some major rewrites and parameterers change to make BeamformIt accessible for the casual user, and to make it less dependent on external libraries (we eliminated dependency with BOOST, libSamplerate and FFTW). As a result, speech researchers currently using previous versions should take care to read the documentation and set parameters correctly so that results are, at least, the same as in previous versions. The most important parameters for speech researchers are "delays_margin" and "skew_margin"/"do_compute_skew". The skew computation should only bother you if you are processing ICSI meetings.
- BeamformIt-3.51.tgz (released July 7th, 2014)
This contains just a few bugfixes found in 3.5. This version is being used inside the AMI recipee for Kaldi. Thanks to Pawel Swietojanski for helping in the debugging and for implementing the AMI recipe for Kaldi, using BeamformIt.
- Github repository (starting November 2015)
As of November 2015 BeamformIt will continue to grop inside Girhub. This will allow for constant improvements to be incorporated into the code. Feel free to fork the project and contribute to it!