Re: Use of PCM and Vorbis in the Mobile Profile of SMIL 3.0

On 8/10/07, Philippe Le Hegaret <plh@w3.org> wrote:
> Requiring the use of PCM and Vorbis in the mobile environment might be
> difficult

The only problem of PCM is the large file size needed to store audio
that way.  Not a good thing on mobile devices, but then again, very
few moble users would be storing PCM audio.

PCM support should be trivial for developers to implemente, even if in
the end it may prove unnecessary to users.

> it *seems* that Vorbis requires more CPU to be decoded than a format
> like AAC, thus rendering its use in a mobile environment more
> problematic.

This is certainly not correct.  Vorbis actually uses less CPU power
than AAC.  The problem is that our optimizations were directed to the
ARM platform, which is not what most mobile platforms use.  We created
a fixed-point decoder called Tremor to solve this, and we are working
on implementing a new Ogg parser (called libogg2), as well as new
floating-point decoder (based on Tremor) to improve the situation for
any kind of mobile device.  Any developer, volunteer or paid, is
welcome to join in the development of these solutions and speed up the
process.

This is the good thing about Vorbis and other open standards like
Speex.  Anyone may comment on them.  Anyone may help fix problems on
them.  Anyone may create a better implementation of them.  This is not
true when it comes to AAC.

> - 3GPP and 3GPP2 already adopted AAC as their audio codec (+ some AMR
> speech codecs). It's hard to imagine the mobile industry willing to
> switch to a different format now.

This is somewhat true.  It does not mean it is a good thing, though.
AAC is a patented technology, with different standards floating around
(e.g. HE-AAC).

The mobile industry, currently, is stuck on MP3, not 3GPP.

> PS: I'm copying Ivo Emanuel Gonçalves and Dominique Hazael-Massieux
> since they might have some input or comments.

Thank you.  I appreciate that.

Best regards,
Ivo Emanuel Gonçalves,
Xiph.Org Foundation

Received on Friday, 10 August 2007 16:05:42 UTC