[Vorbis-dev] Decoding for ambisonic : downmix support

[Richard Lee]

A strategy for downmix support

a)	simple 4.0, 5.0, 6.0 7.0 Speaker Decode on the player via a separate library from the Vorbis core.  This is the easy panning solution on the Wiki Ambisonic Decoder page and requires hardly any computing power.  Most important cos its all about surround init?

This library can be extended to have Shelf Filters & Distance Compensation for more sophisticated players (with more computing power) but still with the standard 4.0 5.0 ... 7.0 layouts as above.

Premium players have a fully programmable Speaker Decoder based on Fon's Linux stuff.

The Vorbis "decoder" defaults to simple 4.0 if the player says its multi-channel but doesn't specify how many.  We should be able to force select the simple 4.0 5.0  ..  7.0 Speaker Decodes do deal with older 'dumb' players.

b)	stereo support.

>simulated blumlein are the simplest downmixes and could be added with a very minimal amount of code. They are also the least satisfactory.

As most of the prize winning recordings made with Soundfield mikes have used them in plain stereo mode, usually Blumlein (eg the Chesky's) I don't think we should look down on simple Blumlein.

The 3 good stereo downmixes are 

- Blumlein fig-8s at 90.  Probably best for loudspeaker stereo
- Back to Back cardioids.  This is best for simple binauaral.
   With Y increased by up to 3 dB good for speakers too.
- UHJ stereo : The most complicated cos long FIRs required.

Each of these will sound better under certain conditions and I would personally want all 3.  But if you asked me which to ditch, the obvious one to throw out is UHJ.

The Vorbis "decoder" defaults to either Blumlein or Back to Back Cardioids if the player doesn't say its multi-channel.


-- 
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.5.441 / Virus Database: 268.18.4/705 - Release Date: 27/02/07 15:24