> -----Original Message-----
> From: Michael Cheng [SMTP:[EMAIL PROTECTED]]
> Sent: den 29 juli 1998 05:23
> To: '[EMAIL PROTECTED]'
> Subject: Re: [MP3 ENCODER] Possible bug in l3psy shortblock
> treshold calculation?
>
> On Tue, 28 Jul 1998, Jan Peman wrote:
>
> > section in l3psy.c:
> > ecb[b] += s3_l[b][k] * eb[k];
> > Shouldn't s3_l be s3_s ?
> Jan,
> didn't you work out that short blocks weren't used anyway?
> At one stage I mucked with the perceptual entropy bit, and
> forced all
> blocks to be long (type 0?). didn't notice any difference in
> quality of
> output.
>
[Jan Peman] uhm.. I think (or at least hope) I said that
you don't notice
any change in quality using only long blocks.. but I think
short blocks are
good (tm) if aiming for good quality.. Maybe it's because
bugs like above
short blocks doesn't give any quality changes.. (just
guessing here..)
Anyway, it's hard to compare quality with sounds that
already are good.
> later
> mike (who has crap ears)
[Jan Peman]
/jp (who probably has even crappier ears)
