[FFmpeg-soc] [soc]: r2688 - aacenc/aacenc.c

Kostya kostya.shishkov at gmail.com
Sun Jul 6 07:38:04 CEST 2008


On Sat, Jul 05, 2008 at 11:17:37PM +0200, Michael Niedermayer wrote:
> On Sat, Jul 05, 2008 at 06:17:12PM +0300, Kostya wrote:
> > On Sat, Jul 05, 2008 at 12:58:16PM +0200, Michael Niedermayer wrote:
> > > On Sat, Jul 05, 2008 at 01:15:20PM +0300, Kostya wrote:
> > > > On Sat, Jul 05, 2008 at 12:11:55PM +0200, kostya wrote:
> > > > > Author: kostya
> > > > > Date: Sat Jul  5 12:11:55 2008
> > > > > New Revision: 2688
> > > > > 
> > > > > Log:
> > > > > Make MDCT output more realistic
> > > > 
> > > > I've tested against libfaac MDCT and seemed to produce
> > > > quite different MDCT spectrum from the same input data.
> > > > After I moved scaling by 512 lavc MDCT began to produce
> > > > almost the same spectrum (without high-freq noise).
> > > > 
> > > > Can somebody tell me if there's a better way to do that
> > > > and obtain the same result?
> > > 
> > > yes
> > > 
> > > 1. go find some cola
> > 
> > Can you send me a couple of thousand liters? Our custom offices definitely
> > need a lot of that stuff too. Oh, and SBU must have your address as well.
> 
> sorry, i need the cola myself :)

What's the current conversion rate from cola liters to hours of sleep?

> > 
> > > 2. think about the difference between short / 512 and float / 512
> > > 3. the else case below is buggy in the same way
> > 
> > Fixed, thanks for pointing at that. Such kind of errors are the nastiest.
> 
> Btw, i think the case of switching dct sizes is wrong as well, it if iam not
> mistaken needs different windows on the right and left side.

It's not wrong, it's absent.
 
> [...]
> -- 
> Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
> 
> If you really think that XML is the answer, then you definitly missunderstood
> the question -- Attila Kinali



More information about the FFmpeg-soc mailing list