>
> Hi all,
> find attached a small tweak to l3psycho_energy (just cut and paste into
> l3psy.c). a 30% improvement in this routine.
> all i did was re-order the checks for chn type.
Thanks, that will be in 3.31, along with the BladeDLL fix from Albert.
I will put 3.31 on the web page after sending this message.
>
> A few ideas/things about this routine
> - when does chn==2 or chn==3? I guess from the code that it's for MID
> and SIDE channels, but as far as i can see, nothing ever calls it with those
> values. Is this code for the future? or is it code from the past that's never
> been removed?
Your right - chn==2 is the mid channel, and chn==3 is the side
channel. I guess some comments about this might be nice :-). The new
mid/side masking thresholds (enabled with -h) computes psycho
acoustics for all 4 channels, L,R,mid & side. It is enabled by the
following block of code in l3psy.c:
numchn=stereo;
if (highq && (info->mode == MPG_MD_JOINT_STEREO)) numchn=4;
for (chn=0; chn<numchn; chn++) {
... psycho acoustics
}
> - why is syncsize=1328 but savebuffer[1344]? The last
> 16 places in savebuffer appear never to be used. not a big point, but just
> wondering if there's some logic there that i'm missing
I think the ISO code used a syncsize of 1344 (which was a bug).
I'll change this.
Mark
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
