Vorbis - Jun 2003 - Calling for 5.1 Mastering experience! (vorbis ambisonics and 5.1)

I've been doing a fair amount of work with Vorbis support for Ambisonics,
which seems to be going along nicely. It seems that there is signifant
interest in coding 5.1 material with Vorbis esp as tarkin becomes more
complete, so I've decided to take a break from pure ambisonic work to look
into this.

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.

To accomplish that I produced a model of an idealized 5.1 speaker
arrangement (with pointsource massless speakers) and a simulated ambisonic
microphone at the listening location. I then measured the ambisonic
impulse response from each speaker and produced convolution filters for
the audio.

The nice thing about this is that you can take the ambisonic decoded
output from vorbis and convolve it with inverse of these filters and get
pretty much the orignal input back or you could also convolve it to match
your own speaker layout and get better results than you would with the
orignal 5.1 material.

My only problem right now is with the LFE (subwoofer channel). I had been
intending to just mix it into the W signal, which I think matches the
intent of the channel. Obviously this really couldn't be recovered on
decode.

My alternative is to use a seperate LFE channel in the vorbis stream,
which is just ugly.

So, what I'm looking for is some guidance about how the LFE is normally
and best used in 5.1 mastering. If I had an infinite supply of perfectly
flat really loud full spectrum speakers and I was setting up an ideal 5.1
setup, would I mix my LFE into my five speakers, or would I grap another
one of those perfect speakers and lay it in the corner with the LFE feed
going into it?

<p><p>--- >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.

Hi,

I am not an expert in this, but my sound engineer friend suggests this
link as a potential answer to your question:

http://www.mkprofessional.com/bass_mgmt.html

Hope this helps,
Ray

On Tue, 2003-06-10 at 13:30, Gregory Maxwell wrote:
> I've been doing a fair amount of work with Vorbis support for
Ambisonics,
> which seems to be going along nicely. It seems that there is signifant
> interest in coding 5.1 material with Vorbis esp as tarkin becomes more
> complete, so I've decided to take a break from pure ambisonic work to
look
> into this.
> 
> 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.
> 
> To accomplish that I produced a model of an idealized 5.1 speaker
> arrangement (with pointsource massless speakers) and a simulated ambisonic
> microphone at the listening location. I then measured the ambisonic
> impulse response from each speaker and produced convolution filters for
> the audio.
> 
> The nice thing about this is that you can take the ambisonic decoded
> output from vorbis and convolve it with inverse of these filters and get
> pretty much the orignal input back or you could also convolve it to match
> your own speaker layout and get better results than you would with the
> orignal 5.1 material.
> 
> My only problem right now is with the LFE (subwoofer channel). I had been
> intending to just mix it into the W signal, which I think matches the
> intent of the channel. Obviously this really couldn't be recovered on
> decode.
> 
> My alternative is to use a seperate LFE channel in the vorbis stream,
> which is just ugly.
> 
> So, what I'm looking for is some guidance about how the LFE is normally
> and best used in 5.1 mastering. If I had an infinite supply of perfectly
> flat really loud full spectrum speakers and I was setting up an ideal 5.1
> setup, would I mix my LFE into my five speakers, or would I grap another
> one of those perfect speakers and lay it in the corner with the LFE feed
> going into it?
> 
> 
> 
> --- >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.
-- 
Ray Heasman <nurf@spamcop.net>

--- >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 Tuesday, June 10, 2003, at 09:30 pm, Gregory Maxwell wrote:
> I've been doing a fair amount of work with Vorbis support for 
> Ambisonics,
> which seems to be going along nicely. It seems that there is signifant
> interest in coding 5.1 material with Vorbis esp as tarkin becomes more
> complete, so I've decided to take a break from pure ambisonic work to 
> look
> into this.
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.

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.

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'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.

Cheers,
  -r

--- >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.

>So, what I'm looking for is some guidance about how the LFE is normally
>and best used in 5.1 mastering. If I had an infinite supply of perfectly
>flat really loud full spectrum speakers and I was setting up an ideal 5.1
>setup, would I mix my LFE into my five speakers, or would I grap another
>one of those perfect speakers and lay it in the corner with the LFE feed
>going into it?
I can't advise too much from a sound design stand point, but I do have a 
fair amount of experience with this on the technology side.
>From what I can tell, the LFE channel is a completely retarded invention. In
speaking with members of the industry who do 5.1 as a business, they seem to 
think it's retarded too.

In your case you would be better off with 5 full spectrum speakers. I think 
the only possible disadvantage would be that low bass coming from 5 speakers 
where you are not sitting in the sweet spot might have them all arriving out 
of phase and muddying your perception. Of course with all the bass coming 
out of the LFE, it can then be out of phase with the other 5 speakers. But 
if it's just high frequency content coming out of the other 5 speakers, it 
probably won't get as muddy as the low bass would.

The following things make the LFE channel stupid:

1) In a standard fold-down, If the 5.1 signal is down mixed to anything 
lower like PLII, stereo, etc., the signal in the LFE channel is completely 
thrown away. So you must not put anything in the signal that is important. 
You can only put stuff in there that's not important. So I start to ask 
what's the point?
2) All receivers perform bass management. If you have pathetic small 
speakers and a kicking subwoofer, your receiver will implement a cross-over 
for you and filter all the bass content out of all channels and send it just 
to the subwoofer. If you have full range speakers it may not send anything 
to the sub woofer and leave the bass content in the mains. So the .1 channel 
is basically emulated half the time anyway.

How this all fits into Ambisonics I no know. You might have to follow along 
and have a .1 channel anyway. I know the .1 channels makes other things hard 
as well, I wish the industry would just dump it (lets go for 8.0 or 10.0 
next, not 7.1 or 8.2!).

_________________________________________________________________
Add photos to your messages with MSN 8. Get 2 months FREE*.  
http://join.msn.com/?page=features/featuredemail

--- >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.

Gregory Maxwell wrote:
> I've been doing a fair amount of work with Vorbis support for
Ambisonics,
> which seems to be going along nicely. It seems that there is signifant
> interest in coding 5.1 material with Vorbis esp as tarkin becomes more
> complete, so I've decided to take a break from pure ambisonic work to
look
> into this.
Great to see there is work done on Vorbis 5.1 !!! Latest AAC encoders 
were slowly taking Vorbis position in the DVD backup scene, because AAC 
5.1 sounds quite nice at 200 - 240 kbps.

About your question :

 From an sound engineering perspective, a separate subwoofer channel is 
more or less useless, it could be muxed fine into the main Left and 
Right speakers, resulting in a 5.0 setup.

Never mux the signal into all channels, especially not the rear 
channels, as those speakers are normally

- not built to be able to generate low frequency stuff at all
- will act against the output of the main speakers, unless you inverse 
the phase by 180° , at least in a finite space = room. If you do 
inverse, to sum up the bass output from rear and front speakers, be sure 
there are a couple of setups where the 'rear' speakers are placed 
somewhere in the room where they found space, but not really in the 
rear, giving you problems again ;-) ...

About 5.1 vs. 5.0 again :

Only in very small rooms, where low frequencies can not even match the 
lambda/2 criterium from one wall to another ( again : finite space = 
room problem ) it could be possible that a subwoofer, placed in one 
corner of the room, could result in better low frequency performance.

Just my 2 cents :)

Christian

<p>--- >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.

Hi all, Hi Gregory,

is there any progress on the surround encoder in Vorbis been done 
recently  ?

Christian

<p>ChristianHJW wrote:
> Gregory Maxwell wrote:
>
>> I've been doing a fair amount of work with Vorbis support for 
>> Ambisonics,
>> which seems to be going along nicely. It seems that there is signifant
>> interest in coding 5.1 material with Vorbis esp as tarkin becomes more
>> complete, so I've decided to take a break from pure ambisonic work
to
>> look
>> into this.
>
>
> Great to see there is work done on Vorbis 5.1 !!! Latest AAC encoders 
> were slowly taking Vorbis position in the DVD backup scene, because 
> AAC 5.1 sounds quite nice at 200 - 240 kbps.
>
> About your question :
>
> From an sound engineering perspective, a separate subwoofer channel is 
> more or less useless, it could be muxed fine into the main Left and 
> Right speakers, resulting in a 5.0 setup.
>
> Never mux the signal into all channels, especially not the rear 
> channels, as those speakers are normally
>
> - not built to be able to generate low frequency stuff at all
> - will act against the output of the main speakers, unless you inverse 
> the phase by 180° , at least in a finite space = room. If you do 
> inverse, to sum up the bass output from rear and front speakers, be 
> sure there are a couple of setups where the 'rear' speakers are
placed
> somewhere in the room where they found space, but not really in the 
> rear, giving you problems again ;-) ...
>
> About 5.1 vs. 5.0 again :
>
> Only in very small rooms, where low frequencies can not even match the 
> lambda/2 criterium from one wall to another ( again : finite space = 
> room problem ) it could be possible that a subwoofer, placed in one 
> corner of the room, could result in better low frequency performance.
>
> Just my 2 cents :)
>
> Christian
>
<p>--- >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-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.