Vorbis dev - Jun 2003 - Calling for 5.1 Mastering experience! (vorbis am bisonics and 5.1)

On 20030610: Gregory Maxwell  wrote:
 (in reply to Ralph Giles)
>> I assume you're aware of the technical documentation on dolby's
site?
>> (http://www.dolby.com/pro/) In particular the surround mixing guide has
>> a lot of detailed guidelines. I don't have any practical experience
>> with it though, so I can't vouch for it.
>Yes I am. I've done a lot of research to determine the 'correct'
speaker
>placement for my ambisonic conversion.
What resources did you use for 'correct' placement?  I *STILL*
haven't
set up my full rig at my new house - and i've been here for eighteen months.
>> In particular it says that the LFE track is for option reinforcement,
>> so while the idea is to encode it separately and mix it into the
>> subwoofer channel, mixing it into the other 5 is acceptable, as is
>> ignoring it. Don't know how safe that is in practice.
>I just can't figure out what the orignal logic is for 5.1 to even HAVE a
>seperate subwoofer channel.  But a lot of the decisions behind the dolby
>5.1 stuff are not mathmatically sound or even intutivly sane.
IIRC it was to use as few bits as possible, w/o distorting the cheapa$$ 
speakers most people throw in the corners behind the couch with bass signal.
(Mustn't take space away from the anti-piracy bits!).

Another guess is, with a choice of only two from this list: 
A|cheap encoding| B|cheap decoding| and C|excellent sound|
Dolby labs picked A and B.  the chips to encode and decode dolby are 
cheap (and sound cheap.)  5.1 is also easy to cookie-cutter a design.

Joe sixpack does not want to 'calculate' where the speaker belongs or 
'program' the delay and phase for its placement.  He just wants a pretty
picture of what the room will look like, and NOT have his wife trip over 
wires.
>> Encoding it separately breaks the symmetry of the ambisonic encoding,
>> but is probably closer to the intent of the original mix. It should
>> compress quite well with it's own codebook given the lowpass, but I
>> think it also tends to share a lot of entropy with the W channel.
>I'd like to think that users of surround Vorbis (including people who
are
>listening to content transcoded from 5.1 sources) will *not* be listening
>on speakers configured per the dolby 5.1 specs but rather listening on an
>(likely irregular) Ambisonic array sized according to their needs and
>resources. Because of this, I'm slightly more concerned with
compatiblity
>with such setups than absoultly perfect mirror of the mix intent.
Bingo - the purist will use an Ambisonic array - but there are many still 
suffering with 5.1
>However, if it's common practice to do evil things with the LFE (say put
>the same LF content thats on the main channels but 180deg out of phase)
>then I will have to seriously consider using a seperate channel.
My kids Disney DVD titles have terrible LFE phase - for instance: Beauty 
and the Beast has the bass in the front channels 180 degrees out of phase
with the 
back channels.  (and all are +/-90 degrees out from the LFE channel to
boot.)
>> > I've decided that the best (from a pure elegance and patent
avoidance)
>> > way
>> > to handle this is to basically decode the 5.1 input into a WXY or
WXYUV
>> > ambisonic signal, which can then be handled by the Vorbis
Ambisonic
>> > support.
>>
>> So U and V are the planar quadupole (m=+/-2) moments? I'm glad to
hear
>> you can get by with the same number of channels.
>
>They happen to fit nicely with the proscribed speaker placements for 5.1.
>I haven't done testing to determine how well I'm really representing
5.1
>yet as doing so will require me setting up a proper 5.1 speaker system
>which seems horribly boring compared to my 14 channel ambisonic rig.
imple'nuff, Plug up you ears and bring your test tracks to your local 
big-box-retail store with a 'demo movie/music room'  Thats how I test 
my encodes for other peoples setups.... and the FutureShop boys just 
might like your "position impossible" soundscapes.

-Myles
--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'vorbis-dev-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

On Wed, 11 Jun 2003, Myles Buckley wrote:
> >Yes I am. I've done a lot of research to determine the
'correct' speaker
> >placement for my ambisonic conversion.
>
> What resources did you use for 'correct' placement?  I *STILL*
haven't
> set up my full rig at my new house - and i've been here for eighteen
months.
I wasn't refering to the ambisonic rig. I was refering to my virtual 5.1
rig.

My ambisonic rig is irregular. There were previously no freely available
good solutions for decoding to an irregular array because of a patent
complicating matters.

However, with the advent of modern brute force computational power, it's
possible to decode an ambisonic signal by measuring the ambisonic impulse
responce for each speaker in the array at the listening positon (this can
even be done without an ambisonic microphone if you don't mind the tedium
of recording seven responses from EACH speaker with an omni reference mic
in seven slightly differnt locations) and then invert and convolve this
response with the ambisonic input signal to give a very good wavefront
restoration at the target location.

This is comuntationally intensive, not at all the simple arithmetic
required for traditional ambisonic decoding.

I plan on releasing a set of tools in the future that will allow you to
input the 3d locations of your speakers and generate such filters without
the tedium of measurment for people who arn't looking for an absolutly
perfect decode.

This method is not encombered by any patents that I'm aware of (as using
direct convolution was impossible before modern computers and software
like brutefir made it possible) and should finally break the ambisonic
restriction of a strictly regular speaker array that made ambisonics
unacceptable for the majority of the users previously.
> >I just can't figure out what the orignal logic is for 5.1 to even
HAVE a
> >seperate subwoofer channel.  But a lot of the decisions behind the
dolby
> >5.1 stuff are not mathmatically sound or even intutivly sane.
>
> IIRC it was to use as few bits as possible, w/o distorting the cheapa$$
> speakers most people throw in the corners behind the couch with bass
signal.
> (Mustn't take space away from the anti-piracy bits!).
Still doesn't parse.. Why not just say that you feed the sub with
everything under 80hz (or whatever the schroder frequency is for the
room) from the other channels.
> Another guess is, with a choice of only two from this list:
> A|cheap encoding| B|cheap decoding| and C|excellent sound|
> Dolby labs picked A and B.  the chips to encode and decode dolby are
> cheap (and sound cheap.)  5.1 is also easy to cookie-cutter a design.
>
> Joe sixpack does not want to 'calculate' where the speaker belongs
or
> 'program' the delay and phase for its placement.  He just wants a
pretty
> picture of what the room will look like, and NOT have his wife trip over
> wires.
Well.. too bad ambisonic mics are so expensive, because it would be
totally possible to build an inexpensive (reletative to okay 5.1 surround
sound hardware) PC based ambisonic decoder that only required 'sane'
speaker placement and holding a mic at a listening location while the box
played and recorded some MLS bursts.
> Bingo - the purist will use an Ambisonic array - but there are many still
> suffering with 5.1
Well with 6 speakers in a ambisonic-ish layout you can do much better
sound.
> >However, if it's common practice to do evil things with the LFE
(say put
> >the same LF content thats on the main channels but 180deg out of phase)
> >then I will have to seriously consider using a seperate channel.
>
> My kids Disney DVD titles have terrible LFE phase - for instance: Beauty
> and the Beast has the bass in the front channels 180 degrees out of phase
> with the
> back channels.  (and all are +/-90 degrees out from the LFE channel to
> boot.)
BAH! I didn't want to hear that.
> simple'nuff, Plug up you ears and bring your test tracks to your local
> big-box-retail store with a 'demo movie/music room'  Thats how I
test
> my encodes for other peoples setups.... and the FutureShop boys just
> might like your "position impossible" soundscapes.
Ha. :) not a bad idea, but what I really want to do is be able to A/B  5.1
playback vs ambisonic simulation of 5.1 playback.

I'm confident that I can make an ambisonic setup produce sound that sounds
like it's coming from each speaker of a 5.1 setup that isn't really
there,
but 5.1 does (occasionally) manage to produce sound that comes from
someplace other than one of it's speakers and I'm concerned that there
may
not be a way to carry that over into an ambisonic encoding since it's
little more than luck when 5.1 ever manages to do that.

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to
'vorbis-dev-request@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.