Re: [Banshee-List] Crossfading and BPM detection

[Brandon Perry <bperry volatile gmail com>]

I don't think you should be holding the BPM detection on such a high pedestal. I am not sure what your intentions to use it are, but I don't see it being more than a neat little feature atm. I think, optimally, there should be some sort of heuristic saying, ok we have all these bpm detections, but there is a 30 second period where all the beats are increased by 30 compared to the rest of the song. Once you decide that, you can use a comma are something to delimit the various bpm's detected throught the song. (Or something to this effect).

On Thu, Feb 19, 2009 at 4:07 PM, Gabriel Burt <gabriel burt gmail com> wrote:

2009/2/19 Neil Loknath <neil loknath gmail com>:

> Bertrand,
>
> I built from SVN trunk last night and played around with the new BPM
> feature.  Unfortunately, my results were disappointing.  I think there could
> be 2 reasons for this:
>
> 1. The GStreamer implementation of BPM detection is not an optimal
> algorithm.
> 2. The sample of music passed to the algorithm is not from the most optimal
> time.  For example, the beginnings of some songs can be very different from
> their middles.

Just fyi, I currently have bpmdetect processing the whole file, and it
spits out bpm values at various points - usually many per file - and I
take the mode of those as the one-true-BPM.

Gabriel

_______________________________________________
banshee-list mailing list
banshee-list gnome org
http://mail.gnome.org/mailman/listinfo/banshee-list

-- 
http://www.volatileminds.net