[FFmpeg-devel] Float vs. fixed-point audio codecs

[Uoti Urpala]

On Fri, 2010-05-14 at 16:20 +0100, M?ns Rullg?rd wrote:
> This situation is undesirable for several reasons:
> 
> - Implementation details should not be reflected in codec names as
>   used on the command line.

> In all of the above cases it may be desirable to enable both versions,
> even if one is faster, letting the user choose.  This brings us to the
> question of how to choose implementation in the API as well as on the
> command line.  One possibility is to introduce a flag which when set
> causes the float version to be chosen, the default depending on

I've considered similar questions in MPlayer. I think your analysis
mixes up two different things that should be distinguished: the user can
specify bitstream type (treat this bitstream as mp3 data instead of
relying on autodetection; but choose whatever method you want for
decoding mp3) or specify a particular decoder (out of 1. lavc internal
int, 2. lavc internal float and 3. external library decoders I want to
use the external library).

For the first case it's better if the bitstream type can be specified
without referring to any particular decoder. But for the latter use it
can be better to have separate decoder names if the differences are
significant enough.

I think it's better if any autodetection logic is not limited to
choosing options within a single already given decoder name (and so
forcing any two alternatives that should be autoselectable to be placed
under the same name), but can also choose which decoder type to use.