Modular Audio Recognition Framework / News: Recent posts

SF.net had an older shell server still up with NFS mounts of project with a bunch of runaway cron jobs that affected MARF's web content (http://marf.sf.net) mount as well. See this for details:

https://sourceforge.net/support/tracker.php?aid=1171512

I was at fault some time ago with a buggy shell script. Now they have fixed the problem, and MARF's web content is regenerated every 15 minutes back again. This includes the CVS tarball, API refs, manual, and other documents updated from the CVS. So, formerly known dead links are no longer dead.... read more

The MARF (Modular Audio Recognition Framework) project is proud to release
a new 0.3.0-devel-20050606 release, at http://marf.sf.net and
https://sourceforge.net/projects/marf/

Release Highlights:

* This release fixes one unfortunate bug in 0.3.0-devel-20040614
that caused the main MARF module to allow selection of one
preprocessing module.

* The new release is much more extensively tested with a grown
number of test applications. It is in fact quite stable despite being
a "-devel" release.... read more

Both MARF and SpeakerIdentApp "experienced" 0.2.1 release due to a small bug in the app, missing files in the previous distro, and somewhat out-of-date doc. files in places. This all believed to be brought back up-to-date with 0.2.1.

The app also includes marf-0.2.1.jar within itself (the -bundle and -bin tarballs) just to make it possible to run it almost "right out of the box" with one download.

MARF 0.2.0 - Our first release after project submission
- new API in MARF - train()/recognize()
- global Makefile build system
- Compilation
- Distros
- TrainingSet improvments
- new modules
- BandpassFilter
- MinkowskiDistance
- More or less functional NeuralNet
- "Random" feature extraction and classification
- MahalanobisDistance, particular case for identity matrix
- improved docs (a lot!)
- code robustness via exceptions and 'final'... read more