[Vorbis-dev] 5.1 surround channel coupling

[Sebastian Olter]

As I mentioned before, I have made some ambisonic encodings and tested them.

2007/1/29, Martin Leese <martin.leese at stanfordalumni.org>:
> I have been doing some work on the Ambsionics
> page at http://wiki.xiph.org/Ambisonics

Your specification of .amb seems to particularly useful, when
considered together with Vorbis - enough to mark Vorbis stream as "AMB
compatible" or just "Ambisonic" and simply follow the spec, using
*number of channels* to determine the type of a soundspace. Thus,
Vorbis can increase in power retaining its current simplicity.

2007/1/31, Sebastian Gesemann <sgeseman at uni-paderborn.de>:
> [ 2 channel coupling revisited ]
> - square polar mapping (with IMHO limited benefit)

A typical movie has most of its sound placed inside the screen (this
not necessarily takes place with music, btw). From the other hand, our
Ambisonic "microphone" sits in center of a soundfield. Thus the X,
"distance" axis duplicates the W ("mono") channel. I have found that
the precision of X axis has significant influence on sound placement
correctness (we ain't got front and back ear), so the coupling mode
should be chosen carefully here.

> [ Possible multichannel extensions for Vorbis ]
>
> (1) Do "pairwise coupling" like it's done with normal stereo material.
> ...
> PROS: should be fairly simple to implement.

But Ambisonics already works =] I have some 130-140 kb/s movie sound
tracks "recorded" ambisonically and encoded with q=0 and they sound
quite good (much better than faac 5.1 at 250-300 kb/s), sometimes it's
hard to distinguish between them and the original soundtrack. The
"recording" takes place in time domain so it can be done even in
oggenc.

> (2) Do "n-way-coupling" (n>2) where more than two channels are used via
> cascaded square polar mapping.

This implies some hierarchy, thus making this especially useful for
Ambisonics, which is hierarchical in its nature (1st order, 2nd order
and so on; even channels in the same order can be considered somewhat
hierarchical, due to their different needs for bandwidth).

> Conclusions:
> Stick with pairwise channel coupling for multichannel signals. I believe
> this is what MPEG2/4 AAC does anyways (me guessing!)

And what about...music? I have some samples encoded from the files
available on Ambisonic Bootlegs. They are live orchestral recordings.
Mic was placed between musicans. Maybe it will be good to listen to
them first, compare with movie tracks, and then decide what to do.

2007/1/31, Ralph Giles <giles at xiph.org>:
> Monty has also cited the lack of (not lossily
> compressed) 5.1 sources to test with as an obstacle here.

Despite 5.1, we got some omnidirectional music, which is definitely
not lossily compressed. They are often 24-bit wav's.

2007/2/1, Ralph Giles <giles at xiph.org>:
> ...the dolby N.M schemes assume a particular speaker
> placement.

Movie sound tracks are often encoded so that sound comes from a
speaker, or at most from between two speakers. This is because most
people place their speakers where they can, not where the spec says.
Thus, there are *several* sound sources in a movie, and as I've
stated, less channels than five can easily mimic the original
soundfield (for example, two hi-fi speakers + light headphones).

> patented

Today's boom of Ambisonics relies on that all patents, except Vienna
decoders for 5.1 (do we need them?) are expired.

> So, while it's an interesting thing to work on, it's a different and
> longer-term project than implementing good support for the 5.1 surround
> mixes most people currently want to encode. The tradeoff becomes more
> attractive when you have 8-12 surround channels, not just 6, and care
> about 3D soundscapes (like in games, art installations, and so on) than
> in film.

A week for an initial support without channel coupling optimizations.
Here's what we got:

1. A well-defined, simple standard for multichannel wave files:
http://www.ambisonicbootlegs.net/Members/mleese/file-format-for-b-format/
Such files can be easily decoded to an arbitrary array of speakers,
even outside Vorbis. The spec handles all issues, thus not requiring
Vorbis stream to be changed except for a single bit of compatibility.

2. An excellent movie player which can do soundfield decoding as well
as to play a movie with an external sound file:
http://www.mplayerhq.hu
Since we will got a standard, the MPlayer's crew will follow it
quickly. Mplayer's support implies FFdshow to be affected, so all
windows players will be compatible soon.

3. Existing Vorbis Surround files (those mine :P) are available. I
will require some kind of hosting directory to upload them, a torrent
tracker is also acceptable. I will also explain how to set up MPlayer
to decode Ambisonics.

Best regards

--
Sebastian Olter