[Webkit-unassigned] [Bug 73545] [GStreamer] FFTFrame implementation

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=73545





--- Comment #10 from Chris Rogers <crogers at google.com>  2012-01-11 11:27:51 PST ---
(In reply to comment #9)
> (In reply to comment #5)
> > Wow, everyone is doing FFTs :) 
> > 
> > Intel also has one for an IPP backend:
> > https://bugs.webkit.org/show_bug.cgi?id=75522
> > 
> > Please see comments in that bug about the layout test which you will want to try as part of validation.
> 
> Hum it doesn't look good :/
> 
> 
> PASS Initial latency of convolver is silent.
> PASS Triangular portion of convolution is correct.
> FAIL First part of tail of convolution is not sufficiently small: -129.7413645172823 dB
> FAIL Test signal was not correctly convolved.
> PASS successfullyParsed is true
> 
> One question, do I need to apply a function window to the timed data before performing the FFT?

No, don't worry about a window.  The FFTConvolver doesn't want any window applied.

Also, don't worry about the failing test.  -129 dB is really quite good and we can adjust the threshold to be less stringent so the test passes.  In fact, the IPP FFT is also adjust this threshold in a similar way if you look at that patch:
https://bugs.webkit.org/show_bug.cgi?id=75522

You can feel free to adjust the layout test so that -129.7413645172823 dB passes

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.