[MPlayer-dev-eng] [PATCH] make ad_libmad output 32-bit integers

[ACS]

----- Original Message ----- 
From: "Nicolas Plourde" <nicolas.plourde at gmail.com>
To: <mplayer-dev-eng at mplayerhq.hu>
Sent: Thursday, June 09, 2005 23:12
Subject: [MPlayer-dev-eng] [PATCH] make ad_libmad output 32-bit integers


>
> On 7-Jun-05, at 10:25 PM, alexander.strange at ithinksw.com wrote:
>>
>> I found it again by searching for my email address in gmane, but it 
>> turns
>> out that I remembered myself doing more work than I actually did,  and 
>> the
>> patch is only to make ad_libmad output 32-bit integers rather than
>> truncating to 16-bit.
>>
>> I like using mp3lib better these days, and the way it generates  floats 
>> and
>> truncates them to 16-bit and then has them converted to floats in  AUHAL
>> and truncated again for the sound hardware certainly isn't optimal,  but
>> I'd have to fix decode_MMX.c and decode_i586.c as well and I've never
>> tried to learn x86 assembly. faad or tremor (or is it going to  switch to
>> ffvorbis?) might be easy enough to do.
>> <MAD32.patch>
>>
>
> Can someone review this patch by Alexander Strange please? Thank you.
>
> ====
> Nicolas Plourde <nicolas.plourde at gmail.com>


there was a dither patch for libmad:
http://mplayerhq.hu/pipermail/mplayer-dev-eng/2005-January/032777.html

I already did  a dither patch for mp3lip & ogg:
http://mplayerhq.hu/pipermail/mplayer-dev-eng/2005-May/034647.html

beware of the mp3lib mmx mode. Since it's broken quality-wise it's no use to 
add dither to it.
However mp3lib decode i_586 mode is fine

Adrian Bacon