[MPlayer-dev-eng] [EPAMP] an effect plugin API for media players: anyone interested?

[Stefano D'Angelo]

Hi,

2008/6/2 Ivan Kalvachev <ikalvachev at gmail.com>:
[...]
> MPlayer have audio filter that is able to load LADSPA plugins, so I
> guess somebody should test your improvements.

Ehm... EPAMP is not improved LADSPA, but whatever ;-)

> We currently don't have maintainer for our audio sound system,
> so it may some time for experts to contact you.

Ok.

> I just took a look of the draft you have given URL.
> I have only 2 remarks.
>
> 1. You define the known audio formats as bit-flags. I see no sense of
> doing that when they can't be combined. I don't expect that all
> possible audio formats would be limited to 32. ;)

>From the draft:

       /* Effects must support one or more PCM sample formats. The pcm_formats
       field is a bitmask that must be set to a combination of PCM sample
       format values (using the bitwise OR operator).

       The host is responsible of passing correct data, according to this
       value, to the run() callback, while it can change format type from
       one call of run() to another without further restrictions. */
    uint32_t pcm_formats;

This field really simplifies the API if you think about it.

Then, ALSA currently supports 47 PCM sample format types (+1 -
unknown), but many of them are rare as gold :-) In case we need more,
we might just use two bitmasks instead of one.

> 2. I couldn't spot where you handle the delay. From what I've heard
> audio filters need to work with past and future samples, so some
> filter add additional delay. Using few such filters could accumulate
> significant delay that could be noticed in application that need to
> keep sync between audio and video.

No delay handling actually, and thanks for having brought this kind of
problem up (I would have never thoguht about it).
I mean, if the host knows about how much delay an effect introduces,
when using a non real-time audio source, it can be balanced.

Ok, this will be in the next draft ;-)

> I mean no offense, but our own audio system seems more mature than
> your draft And I think that our audio system have a lot of space for
> improvement.
> (well, i guess it's normal for a draft;).

No offense taken at all and yes, this is just a draft at the moment ;-)

Cheers,

Stefano