[Clam-devel] Turnaround - Chord Extractor application

[David García Garzón]

On Dijous 17 Juliol 2008, Pawel Bartkiewicz wrote:
> > Notice that binary files (mainly pngs) are missing in the patch (svn diff
> > doesn't work with binaries). In any case, please, be conservative and
> > include
> > just those images you are actually using. Not those that you will
> > eventually use. If you resend the patch (and the missing images as
> > attatchment or in a tarball) i will commit it.
>
> They were copied from Annotator/src/images:
> fileopen.png
> exit.png
> player_pause.png
> player_play.png
> player_stop.png

Ok, but then, we should adapt the image.qrc in order to include just those 
images. Change it, resend the patch and i will copy the files and commit it 
all.

> > Also the monitor is something you are adding now because you are doing
> > realtime visualization. As long as you will remove the realtime
> > visualization
> > the monitor must be removed from Turnaround.
>
> I was trying to write my own DataSource today but the ChordExtractor
> code (from Annotator) is messy indeed. Can I use TonalAnalysis
> instead? It seems to be a bit cleaner way.

You are right. I just thought that tonaAnalysis still had no segmentation, but 
Roman Goj added it at the end of his GSoC. It might have some bugs (frame 
location and closure bugs mostly) but we should be just carefull with that. I 
just did some fixes right now.

I would like to have Pools as data storage in order to share code with 
annotator anyway. Notice that current PoolDataSources are too coupled to 
Annotator data structures such as the Project. We could build one from 
scratch and them make them converge later.

> > I thought that was clear, but let me
> > remind you just in case, the procedure is like that: On opening a file,
> > the file is analysed offline. That means:
> > * that you can use the FreeWheelingPlayer for that (no Jack,
> > PortAudio...). * a progress bar using the progress control would be
> > perfect here * the results of such an analysis, are not visualized but
> > stored in an in-memory storage (we call it DataPool) as ChordExtractor
> > does before dumping.
> > At that point the user may play and seek the song using and
> > AudioFileMemoryLoader connected to an AudioSink. The visualizations will
> > take
> > the data from a DataSource that access the DataPool, but selecting the
> > analysis data at the time position sent by the AudioFileMemoryLoader
> > output control.
>
> That's how I'm trying to do it. However, I don't know how to use the
> DataPool, because the ChordExtractor from Annotator uses it for
> dumping the data to a file, which isn't my case.

Yes, the ChordExtractor dumps the pool it into a file, and the Annotator 
rebuilds the pool from that file, so once you have the pool you have it. It 
is the end point of the analysis and the starting point for the 
visualization. No need to store.

> > Wow. Perfect. An IRC session would be perfect for that. Are you still
> > experiencing network problems?
>
> I will be most likely online tomorrow evening. It'll be Friday,
> though, so maybe you prefer any other time?

I'll be online until 18h (spain)

-- 
David García Garzón
(Work) dgarcia at iua dot upf anotherdot es
http://www.iua.upf.edu/~dgarcia
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 189 bytes
Desc: This is a digitally signed message part.
URL: <http://lists.clam-project.org/pipermail/clam-devel-clam-project.org/attachments/20080718/c67702e1/attachment.sig>