Subtitling a video/song is a tedious task, not only one has to write the subtitle, but also one has to specify its timing (start and end times). ASTA project tries to automatically determine the subtitles timing based on the, possibly polyphonic, audio input. We found Marsyas to be the most suitable tool for both signal processing (i.e. feature extraction) and machine learning (i.e. training and classification). Beside its efficiency, it provides such a complete solution to audio-analysis that we didn't need any other library. We thank Marsyas team for open-sourcing such a great project.

http://sourceforge.net/projects/asta-annotator/

Oh wow, that looks pretty sweet, I can't even subtitle stuff anyway cos I can't hear anyway but that's pretty sweet.

I'll download and test this one out. I'll let you know what I think.

Tom

edited to add:

Okay, so to run it, you need to compile it under linux. From the helpfile:

Code:

Currently works on Linux platforms with cmake, ffmpeg & Python. After download
       $ cd /path/to/asta/machine-learning/tools/marsyas-xx/src
       $ cmake .
       $ make

I don't have any plans to do this, so I won't be reviewing it. Let me know if you plan on distributing a binary that can be run on Windows, or a linux virtual machine image with ASTA loaded and ready to go.
Good luck!

Tom.