[MPlayer-dev-eng] [PATCH] support floats in ad_libmad

[Alexander Strange]

On Mar 10, 2005, at 1:13 AM, D Richard Felker III wrote:

> On Thu, Mar 10, 2005 at 12:26:36AM -0500, Alexander Strange wrote:
>> This patch allows libmad to output floats with -af-adv force=4+ 
>> instead
>> of truncating everything to 16-bit.
>
> Not acceptable since float is not the native format. Instead make it
> output 24bit or 32bit or whatever the native format is and add support
> for this to libaf.. Conversion to float does not belong in a decoder.
>
> Rich

The default format is a 32-bit integer shifted down three bits to allow 
for overflow error. The attached patch will make it output signed 
32-bit integer; leaving all conversions completely out of the decoder 
would need a volume filter inserted after it, and I don't know enough 
about the setup to do that. Should it do that?

Other things I notice: ads faad, libvorbis, mp3lib, and liba52 (unless 
-af-adv force=4 is used, and then it still downmixes surround) don't do 
this right. Also, if resample is used, it will continually complain "8 
bytes of audio data lost due to buffer overflow, len = 8".

-------------- next part --------------
A non-text attachment was scrubbed...
Name: MAD32.patch
Type: application/octet-stream
Size: 2385 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/attachments/20050310/02e261e1/attachment.obj>
-------------- next part --------------