[FFmpeg-devel] [PATCH] G.729 and G.729D decoders

[Vladimir Voroshilov]

On Mon, Apr 21, 2008 at 12:24 AM, Michael Niedermayer <michaelni at gmx.at> wrote:
>
> On Mon, Apr 21, 2008 at 12:11:16AM +0700, Vladimir Voroshilov wrote:

[...]

>  >
>  > Yes, LP decoding in soc/amr looks very similar to mine.
>  > How those can be joined?
>  > What about creating new file (celp.c for example) ?
>
>  This was exactly my idea (celp.c, lsp.c, ... whatever names makes sense, and
>  more files with less content are better than few big files)

currently selected acelp_lpc.* and acelp_math.*

>  > Quick look shows me that lsf2lsp, reorder_lsp can be merged.
>  >

More precise look give a trouble: amr code uses "int" type everywhere, while my
code uses "int" only in necessary places. So common routines (mine and
amr's) differs
only in used types.
I'm not sure that i can correctly replace int with int16_t in amr
code. (superdump?)

>  > decode_*_pulses_* routines can be merged too by using lookup tables in amr
>  > instead of hardcoded shifts  and multiplications (see mine
>  > decode_fc_vector, unified
>  > routine in cost of several additional lookup tables).
>  >
>  > What patches should i prepare?
>  > One for celp.c and one for amr soc project in the same mail (in
>  > separate thread, of course)?
>
>  ok
>

I'm going to post incorrect patch (with ignorance of "int"<>"int16_t")
for amr (just for starting discussion on). ok?


-- 
Regards,
Vladimir Voroshilov mailto:voroshil at gmail.com
JID: voroshil at gmail.com, voroshil at jabber.ru
ICQ: 95587719