Dave Malham wrote:
Hi Stefan,
I doubt if B+ would meet the currently perceived needs of cinema
surround mixers/producers since it does not have the ability to go
"discrete". B++ might be enough - that's first order + 5.1 (I just
made that up :-)). A better option would be at least third order,
preferably fifth - to get the most bang for buck - but to fit in the
22.2 channels that seem to be coming over the horizon fast, maybe
fourth would work better in terms of channel count.
Just some further ideas, open for some discussion, and meant to obtain
some possible .AMB or HOA based 3D audio standard... ;-)
- For cinema 3D audio, you certainly could apply some mixed order
approach, using a lower vertical order than horizontal order. You save
a lot of channels...
Without mixed orders: 4th order needs 25 channels, above 22.2 channel count.
However: Current cinema 3D layouts don't use more than 3 levels! This
would imply that you could use 2nd order for the vertical components,
because any higher (vertical) order can't be resolved.
Even this might be overkill:
a) Dolby Atmos seems to have only 2 levels.
b) 22.2 has three levels, but one is "below surface". Also in this
case, a 1st order vertical component looks to be sufficient.
(Negative elevation, "zero elevation", upper ring.)
c) Auro-3D has three "positive" levels. But the highest level is just
some over-the-head/"voice of god" loudspeaker. Again, 1st order for the
vertical component(s) seems probably adequate.
-----------------------------------------------------------------------------------------
Now, if 2nd order "vertical component" is enough FAPP:
You easily fit even 6th order horzinal / 2nd order vertical "B format"
(doesn't exist, ok) into 22.2 channels.
13 (horizontal 6th order) + 4 (vertical 2nd order) + (optional) up to 5
discreet front channels + 2 LFE channels fit into... 21.2. (Fill up one
channel with silence, if this mattered... I won't claim any patent for
this decisive idea! :-) )
("Naive" 4th order was 25 channels, so probably 27 because of two
additional LFE channels...)
Of course I was planning to use some mixed order approach anyway, but
here anybody could see how channels can be saved. (You can cut channels
if nobody can hear them after decoding. Or let us assume that there is
no advantage in using components for which the loudspeaker
configuration would be underspecified... If you < can > hear a
difference, the result of higher orders decoded to some undespecified
array should be worse, not better.)
I don't think that any .AMB or HOA format would have to fit into 22.2,
anyway... I just wanted to demonstrate that you could use even 6th order
within a still acceptable (global) channel number count. If we are
admitting that all existing cinema 3D audio layouts will have about 2-3
levels. Maybe 4, but not more.
Anyway, my current "standard proposal" would (still) be to use the
"classical" 3rd/2nd mixed order .AMB format, extended with 2 LFE
channels and (optionally) 3-5 front channels (direct channels).
This is about 13-16 channels, which seems to be ok. (For home and
"mobile" use, 3rd oder horizontal/1st order vertical fits into 8
channels, if the 8-channel limit still matters.)
Up for discussion:
- The sweet spot of any soundfield based approach for cinema audio
should be equal or superior to 5.1. Can 3rd order .AMB deliver this
requirement?
(To compare, it might make sense to add the 3 to 5 "direct" front
speakers, as I have proposed before. B format would refer to a 3rd order
or mixed order - say 3rd/2nd order - soundfield. Compared to the
original B+ format, we also have now 3 or 5 direct front channels, and 2
LFE channels.)
- Could you further improve the decoding techniques for 2nd/3rd order
soundfields, based on perceptional ideas (like 1st order
Ambisonics/FOA), and/or via blind source separation algorithms (i.e.
Harpex)?
- If so, do we really need anything above 3rd order? (But 4th, 5th and
6th order can be done, even within 22.2 channels. Which has been
demonstrated before,)
Best regards,
Stefan
P.S.: To my best knowledge, it is not so clear how the best perceptional
decoding strategies for 2nd or 3rd order Ambisonics should look like.
Feedback welcome...
_______________________________________________
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound
