On Thu, 14 Oct 1999 [EMAIL PROTECTED] wrote:
> hey all,
> just tidied up the source a bit, renamed some things for easier viewing
> and released toolame01d (optimized layerII) code. www.cryogen.com/mikecheng
> Anyone who wants a break from hardcore LAME coding is welcome to chill
> to a bit of layerII :)
> Patrick has a few ideas floating around, and I'll add his new
> scale_factor_calc on the weekend (fingers crossed)
>
About "my" ideas; I still have to have a more detailed look during the
weekend, but it seems that majority of my former suggestions have been
incorporated into Takehiro's snapshot (sorry for poluting the list
before and now); is this going to be put into "official" lame, and
independantly of that, into toolame (who has plans, when; maybe I can
help)?
Possible improvement remarks (against Takehiro's snapshot):
- add sblimit / "cut off" to stop filtering subbands/frequencies which are
dropped afterwards anyway ??
- Takehiro's mdct.c; window_subband;
why did he/you move away from pointers, aren't arrays less efficient
i.e. is win[i+16] not compiled less efficient than *ep0++
And when using pointers why not different y[i] loops;
>> see http://telin.rug.ac.be/~pds/mpeg_audio/xlame/here.html
Of course then you would need a lot of pointers, but still ??
regards,
Patrick.
--
MP3 ENCODER mailing list ( http://geek.rcc.se/mp3encoder/ )
