Hiya all, (mostly John33)

Now that MPC is open source(on corecodec.org or something) SV7+SV8

how about doing an optermized build of SV7?? IC7 I think (like the others on rarewares)

i know it wouldn't inprove quality but it should give it a speed boost!!

any idea's??

Thanks all Lemonzest

I'll grab the source and check it over.

thanks, hope i'm not much of a pain



Lemonzest

Anyone who attempts to make a faster compile should by all means check if the output is bit-exact identical to that of the official one. We've had problems there with LAME in the past. If it's not identical and people still want to use it, it would basically need new listening tests for integrity.

i was thinking that too, maybe if/when john33 is finished we could compress some known samples with both compiles and md5/crc the output. i would be willing to help.

thanks for reminding me, Lemonzest

I would prefer an "official" binary , instead of many various, sometimes obscure versions of mppenc.
It'll make many things easier, like comparing results, as it is assumed that different options during compilation could lead to different results.These differences may not be abxable but the point remains.
And some compilation options may just bork completely the file or the bitstream.

Agreed. The fact the MPC encoder has been closed source for a while has been good for the format in a small way - you can be reasonably sure an mpc file is going to work. Official binaries are the way forward.

Isn't mppenc very fast anyway? I can't remember having any problems with encoding speed when I used it. Well, opinions may differ

Well, the Intel compilers seem to break the encoder, so that kind of reduces the options anyway!!

Same here. Encoding a tune with standard using an ICL7.1 build ended up at 88kbps and absurdly horrendous artifacts.

We will probably release official builds when the new code is ready. Frank is preparing a 1.2x version and we will start release from that point (I'll use MSVC7). Any problems with other compilers should be investigated, IMO.

BTW, I've tried 1.15r under OSX using GCC 3.3 (Apple XCode) and both mppenc and mppdec handle files incorrectly. So I assume there are other compiler dependencies here and there...

I noticed a version of 1.15r for OSX compiled by Dibrom on RareWares ? How did he do ? The sources as they are in CVS won't compile on OSX out of the box.

He didn't get the sources from the CVS. I think Case sent them to him after Frank's approval.

Its sort of amazing. After lots of huffing, puffing, pleading and outright whining things are happening again with MPC.

But it was most probably the same sources. Anyway, I PMed Dibrom to have more info about it.

hi,

i am trying to develop a small Win32 application which incorporates MP3 decoding/encoding and ability to play audio streams (like this one http://switch5.castup.net/_custom/radiobu99fm/audio.asx WMV ?)
one of the limitations that i have to keep binary as small as possible. preferably under 1M.
is there any documentation available on the source code of the MPC ?
what module i should to checkout from the official CVS at SF.net ?
any compilation instruction ?

i appreciate any help