[gst-devel] Move of gst-plugins-bad/gst/spectrum

Stefan Kost ensonic at hora-obscura.de
Sat May 27 08:31:06 CEST 2006


more thoughts related to the design of anayser plugins:

Using the GstLevel like approach has the benefit that one can just
insert the analyzer like this:
   element ! level ! element
The spectrum element needs to be used like this
   element ! tee name=t ! element
                     .t ! spectrum ! fakesink
In thsi case we need fakesink (use the handoff-signal to grab buffers)
and an additional tee.
Another advantage of the GstLevel approach is that one can chain
analyzers. Imagine having a full audio-analysis module, where this gets
   element ! tee name=t ! element
                     .t ! level ! spectrum ! panorama ! ... ! fakesink
This way one could plug the analyzers on demand (put then in a bin with
a sink-ghostpad) and use the element messages to get the results.

If noone comes up with a good reason against the gstLevel approach I am
going to rewrite the Spectrum plugin to go the same way.


Am Montag, den 22.05.2006, 09:07 +0200 schrieb ensonic:
> hi,
> over the weekend I have sort of recovered the spectrum analyzer in
> gst-plugins-bad/gst/spectrum. I belive it fullfills the requirements of the
> 'plugin-moving' policy. Still there are some open 'issues' I'd like to
> discuss before:
> 1) where can it go: ugly/good?
> The plugin contains integer FFT routines from xmms (GPL). As we link to
> them the plugins becomes GPL too. The xmms-files miss the copyright
> headers. In the current cvs the files don't seem to be present anymore (or
> in this form):
> http://cvs.xmms.org/cvsweb.cgi/xmms/xmms/fft.c?rev=1.3&content-type=text/x-cvsweb-markup
> 2) delivering results
> The resulting spectrum is pushed out as a buffer right now. If one asks for
> 256 frequency bands, the resulting buffer is guint[bands] with the level
> per band. This basically works. What I am worried about is the caps to use
> for such data?
> And there is an alternative. The level element works as a basetransform,
> which leaves the data as it is, but sends messages with analysis results. I
> could turn the spectrum element into a subclass of basetransform and do the
> same. Would that be better?
> 3) code review
> Who would support moving the plugin. Can anyone review the changes?
> http://webcvs.freedesktop.org/gstreamer/gst-plugins-bad/gst/spectrum/
> Thanks!
>   Stefan
> -------------------------------------------------------
> Using Tomcat but need to do more? Need to support web services, security?
> Get stuff done quickly with pre-integrated technology to make your job easier
> Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
> _______________________________________________
> gstreamer-devel mailing list
> gstreamer-devel at lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/gstreamer-devel

More information about the gstreamer-devel mailing list