[pulseaudio-discuss] PulseAudio Updates

[Lennart Poettering]

Hi!

It has been a while since I last posted on this mailing list. A quick
update on the current situation of PulseAudio: 

I am still working on my thesis and will need another month before I
can focus on PulseAudio full-time again. In the meantime Pierre Ossman
agreed to maintain PulseAudio. Please note however, that he's not
working full-time on it, nor is he integrating new stuff.

In January I've been attending the FOMS and linux.conf.au
conferences. On LCA I did a presentation about PulseAudio, which you can view
online on:

http://mirror.linux.org.au/pub/linux.conf.au/2007/video/talks/211.ogg (Theora)

The slides:
http://0pointer.de/public/pulseaudio-presentation-lca2007.pdf

During FOMS we discussed the current state of Linux audio. After a
long arguing we agreed that introducing a new abstracted
cross-platform audio API (dubbed "libsydney" for now) is a good idea,
to work against the current mess of Linux audio systems and
APIs. Yepp, trying to fix that we have too many incompatible,
competing audio APIs and systems by introducing yet another one seems
paradoxical, but we think this is the best way to go.

As many of you might know I am not happy with the current PulseAudio
audio streaming API. While it is very powerful it is also very
complicated due to its asynchronous nature. 

It is my intention to adopt this new abstracted audio API as the only
official audio playback API for PulseAudio. That way, if you program
against PulseAudio you get cross-platform support for free.

You might ask why we decided to design our own new API instead of
adopting an existing API such as PortAudio. The most important reason
here is that all the current APIs are not designed with sound servers
in mind. The new libsydney API in contrast exposes a buffering model
that is much more suitable for networked sound servers. Besides
working better on these setups the API makes certains things much
easier to implement. On normal hardware soundcards the buffering model
is emulated. Basically the new API is based on the PulseAudio API, but
without the asynchronicity and cleaned up for non-PulseAudio
backends. The API design has been done by Jean-Marc Valin
(Xiph.org/speex), Mikko Leppanen (Nokia) and me.

Besides a draft header file no code has been written yet. Hence enough
of these fluffy promises for now. I will post more about this
on my blog eventually. 

Another outcome of FOMS is that we now have - thanks to Jean-Marc
Valin - an LGPL fixed-point resampler implementation (part of libspeex
SVN), which will hopefully speed up resampling in PulseAudio (such as
the one done by module-combine) quite a bit.

Lennart

-- 
Lennart Poettering; lennart [at] poettering [dot] net
ICQ# 11060553; GPG 0x1A015CC4; http://0pointer.net/lennart/