#27 interleaved 32-bit audio sample processing

[Michael Fink]

Hi,
I'm using LAME as part of winLAME. For now I use the lame_encode_buffer_interleaved() function that takes 16-bit interleaved samples, but I also want to support 32-bit interleaved sample processing, and so I came up with a patch to implement a lame_encode_buffer_interleaved_int() function to put 32-bit signed interleaved audio samples into LAME. I hope this can be put into LAME, since it's only an extra function, not a change of existing functionality. Thanks in advance!

[Alexander Leidinger]

Logged In: YES
user_id=90585

An int may be 64bit on 64bit systems (e.g. Athlon 64). I
suggest to use a 32bit type like int32_t instead.

[Michael Fink]

Logged In: YES
user_id=124874

The function lame_encode_buffer_int also uses int as data
type for the samples and would suffer from the same error. I
don't want need to depend that int is a 32-bit value, but I
want to put samples into LAME that come interleaved "int"
values. So I consider the patch ok in that sense.

[gregTelevirtual]

Logged In: YES
user_id=1612121

Hi,

this isn't really gonna help you guys, but from reading
this i think you guys may be able to help me, as you know
about the things i'm having trouble with.

The pcm I'm creating is a mono file, I believe that to get
around this issue I need to copy my buffer so I have two
channels, left and right and feed them into
lame_encode_buffer() the main problem is however, using
Microsoft visual Studio. NET 2003 I get link error 2019:
unresolved symbol 'lame_encode_buffer()' reference in
function 'Our sound encode function'. I've tried using
all the other lame_encode functions and are having to make
do with lame_encode_buffer_interleaved() but this appears
to half the quality of the file output. In short, we need
to know a) the best way to encode a mono pcm file and b)if
this method doesn't use the interleaved function what
might cause the rest of them to generate the link error.

sorry for hijacking your thread but i'm in desperate need
of some help.

regards

Greg

[Michael Fink]

updated patch for lame.c to add function lame_encode_buffer_interleaved_int32_t()

[Michael Fink]

Logged In: YES
user_id=124874
Originator: YES

I updated the patch for the latest CVS HEAD version, lame.c revision 1.319. I also changed the type from int to int32_t as suggested. I hope the patch can be applied to CVS. Many thanks!
File Added: lame.c-lame_encode_buffer_interleaved_int32_t.patch

[Michael Fink]

updated patch for lame.h to add function lame_encode_buffer_interleaved_int32_t()

[Michael Fink]

Logged In: YES
user_id=124874
Originator: YES

File Added: lame.h-lame_encode_buffer_interleaved_int32_t.patch

[Michael Fink]

Hi,
This patch is the only change to LAME that is needed so that I can use the libmp3lame.dll that can be compiled from source or that can be downloaded from Rarewares. After 6 years I got the latest LAME sources, re-applied the changes and tested it with winLAME. The result is that mp3 to mp3 (and other conversions) can use 32-bit samples, losing fewer bits than using 16-bit short int samples. I hope the patch can go into LAME at some point. Thanks!
Michael

[Alexander Leidinger]

committed to CVS