Re: [Sursound] octofile release

[Martin Leese]

Aaron Heller wrote:

> There's a IETF proposal from folks at Google for "Ambisonics in an Ogg Opus
> Container", based on
>
>   Nachbar, et al., Ambix - A Suggested Ambisonics Format. 3rd International
> Symposium on Ambisonics and Spherical Acoustics, Lexington, KY (2011)
>
> and the idea of a default stereo decode from Etienne Deleflie's Universal
> Ambisonic work

I would instead recommend my proposal for a
stereo decode described at:
http://members.tripod.com/martin_leese/Audio/stereo_mix_proposal.html
http://members.tripod.com/martin_leese/Audio/StereoMix_chunk.html

This can be used not only in Ambisonic files
(see Example 3 on the Stereo Mix chunk page),
but also in non-Ambisonic multi-channel files.
This makes it more likely to be adopted.

Regards,
Martin
-- 
Martin J Leese
E-mail: martin.leese  stanfordalumni.org
Web: http://members.tripod.com/martin_leese/
___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.

Re: [Sursound] octofile release

[Martin Leese]

Marc Lavall?e wrote:
...
> I don't know if the fine Xiph developers can "just" extend the
> definitions of FLAC, if a special Ambisonics mode would be required, and
> to what extent the 8-channel limit (as a limit) is a political issue.

Extending a FLAC stream beyond eight
channels is not possible.  There are two
problems; one simple, one less so:

The simple problem is that the field in the
header used for the number of channels is
only three bits.

Perhaps comparing FLAC with the Ogg
container and Vorbis codec will aid
understanding of the more difficult problem.

With Ogg, different streams can be either
chained (sequential) or grouped
(parallel/interleaved).  Typically, metadata
streams would be chained (so they appear
before any audio data) and audio streams
would be grouped.

Within a single FLAC stream the audio is
split into blocks which are grouped.  But within
each block the eight channels are chained.
This makes sense with a maximum of only
eight channels.  Within a Vorbis stream the
audio is split into frames which are grouped.
However, because a Vorbis stream can
contain up to 256 channels, within each frame
the channels are also grouped.

So the maximum of eight channels is really
embedded into the FLAC standard.  To change
this would require a whole new standard (or
the use of multiple grouped FLAC streams in
an Ogg container).

Regards,
Martin
-- 
Martin J Leese
E-mail: martin.leese  stanfordalumni.org
Web: http://members.tripod.com/martin_leese/
___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.

Re: [Sursound] octofile release

[Aaron Heller]

There's a IETF proposal from folks at Google for "Ambisonics in an Ogg Opus
Container", based on

  Nachbar, et al., Ambix - A Suggested Ambisonics Format. 3rd International
Symposium on Ambisonics and Spherical Acoustics, Lexington, KY (2011)

and the idea of a default stereo decode from Etienne Deleflie's Universal
Ambisonic work



   https://tools.ietf.org/html/draft-ietf-codec-ambisonics-07

Martin Leese has posted pointers to it here from time to time. Early
versions had errors, which I informed them of and were fixed in later
versions.


Aaron Heller (hel...@ai.sri.com)
Menlo Park, CA



On Sun, Jul 29, 2018 at 4:45 PM, Stefan Schreiber 
wrote:
>
> Short comments (to text below):
> (1)
>
> The ambisonics input channels can’t be coded in some 7.1 (channel
coupling, LFE) style, agreed.
>
> With opus you seem to need channel mapping #255, not #1 - the latter
 corresponds to (classical) 2D surround sound layouts.
>
> (2) I believe that  FB is already using up to 11 channels coded with
opus, although I am not absolutely sure about. (Could say more about this
in a few weeks, hopefully.)
>
> (3) Multi-channel != (classical) 5.1/7.1 surround sound.  In fact
surround sound is not a synonym for 5.1/7.1.
>
> We are on the surround list, so should know about this!
>
> But now we are getting slightly confused: Xipg.org’s Opus or Flac don’t
have to care about Dolby Digital or DD+ 5.1/7.1, right? So maybe you mean
5.1/7.1 channel mapping?
>
> To claim that 5.1 is “Dolby” doesn’t make sense. (There is an official
ITU layout standard, and many versions implemented/defined by DTS, Mpeg,
Sony and “anybody else”.)
>
>
https://www.itu.int/dms_pubrec/itu-r/rec/bs/R-REC-BS.775-3-201208-I!!PDF-E.pdf
>
> So (righteously) you can implement 5.1 and 7.1 audio tracks since
Vorbis...
>
> (4) A lossless compression format could be used for mastering. I meant
this.
>
> Best,
>
> Stefan
>
> - - - - - - -
>
> Citando Marc Lavallée :
>
>> Le 2018-07-29 à 05:56 PM, Stefan Schreiber a écrit :
>>
>>> 1. I believe that the opus encoders/decoders have always supported more
than 8 channels.
>>
>>
>> Correct, but when encoding 8 or less channels, correlation is applied in
ways that are incompatible with Ambisonics; for example, the LFE channel is
filtered... With more than 8 channels, Opus don't correlate channels, but
it does now if the input stream is Ambisonics (and if the Ambisonics mode,
disabled by default, is compiled in).
>>
>>> 2. The next question is what ogg channel mapping and consequently
real-world browsers allow...
>>>
>>>
>>>
>>>  But in some sense the hack you did is known. (More complicated is
maybe to make it work...)
>>
>>
>> I tried only with 4 channels. It worked. I don't know if browsers are
now capable to support more than 8 channels. If the Octomic is getting
popular with VR content producers, maybe browsers will start supporting
streams with more than 8 channels (without systematically down-mixing them
to stereo).
>>
>>> 3. If they already plan to issue some ogg ambisonics standard (using
ogg opus of course) since at least 2016: You also need an associated
mastering standard, which would not change or compress any audio data.
Correct?
>>>
>>>
>>>
>>>  So what is “political” about extending the channel count of FLAC?
>>
>>
>> Multi-channel still mean Dolby 5.1 or 7.1. There's an inertia because
"standards" were designed as vendor lock-ins.
>>
>>> Compromise proposal:
>>>
>>>
>>>
>>>  4. So let’s maybe use .wav or .caf for the “mastering format”.
Microsoft and Apple already allow more than 8 channels...
>>
>>
>> Sure. Lossy codecs are not suited for mastering.
>>
>>> P.S.: “Joint stereo” you could classify as parametric coding.
>>
>>
>> Ok.
>>
>>
>>
>>  Marc
-- next part --
An HTML attachment was scrubbed...
URL: 

___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.

Re: [Sursound] octofile release

[Stefan Schreiber]

Short comments (to text below):
(1)

The ambisonics input channels can’t be coded in some 7.1 (channel  
coupling, LFE) style, agreed.


With opus you seem to need channel mapping #255, not #1 - the latter  
 corresponds to (classical) 2D surround sound layouts.


(2) I believe that  FB is already using up to 11 channels coded with  
opus, although I am not absolutely sure about. (Could say more about  
this in a few weeks, hopefully.)


(3) Multi-channel != (classical) 5.1/7.1 surround sound.  In fact  
surround sound is not a synonym for 5.1/7.1.


We are on the surround list, so should know about this!

But now we are getting slightly confused: Xipg.org’s Opus or Flac  
don’t have to care about Dolby Digital or DD+ 5.1/7.1, right? So maybe  
you mean 5.1/7.1 channel mapping?


To claim that 5.1 is “Dolby” doesn’t make sense. (There is an official  
ITU layout standard, and many versions implemented/defined by DTS,  
Mpeg, Sony and “anybody else”.)


https://www.itu.int/dms_pubrec/itu-r/rec/bs/R-REC-BS.775-3-201208-I!!PDF-E.pdf

So (righteously) you can implement 5.1 and 7.1 audio tracks since Vorbis...

(4) A lossless compression format could be used for mastering. I meant this.

Best,

Stefan

- - - - - - - 

Citando Marc Lavallée :


Le 2018-07-29 à 05:56 PM, Stefan Schreiber a écrit :

1. I believe that the opus encoders/decoders have always supported  
more than 8 channels.


Correct, but when encoding 8 or less channels, correlation is  
applied in ways that are incompatible with Ambisonics; for example,  
the LFE channel is filtered... With more than 8 channels, Opus don't  
correlate channels, but it does now if the input stream is  
Ambisonics (and if the Ambisonics mode, disabled by default, is  
compiled in).


2. The next question is what ogg channel mapping and consequently  
real-world browsers allow...




 But in some sense the hack you did is known. (More complicated is  
maybe to make it work...)


I tried only with 4 channels. It worked. I don't know if browsers  
are now capable to support more than 8 channels. If the Octomic is  
getting popular with VR content producers, maybe browsers will start  
supporting streams with more than 8 channels (without systematically  
down-mixing them to stereo).


3. If they already plan to issue some ogg ambisonics standard  
(using ogg opus of course) since at least 2016: You also need an  
associated mastering standard, which would not change or compress  
any audio data. Correct?




 So what is “political” about extending the channel count of FLAC?


Multi-channel still mean Dolby 5.1 or 7.1. There's an inertia  
because "standards" were designed as vendor lock-ins.



Compromise proposal:



 4. So let’s maybe use .wav or .caf for the “mastering format”.  
Microsoft and Apple already allow more than 8 channels... 樂


Sure. Lossy codecs are not suited for mastering.


P.S.: “Joint stereo” you could classify as parametric coding.


Ok.



 Marc

-- next part --
An HTML attachment was scrubbed...
URL: 

___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.

Re: [Sursound] octofile release

[Stefan Schreiber]

Citando Marc Lavallée :


Le 2018-07-29 à 03:08 PM, Stefan Schreiber a écrit :


Citando Marc Lavallée :


Excerpt from https://xiph.org/flac/faq.html#general__channels :

  "FLAC supports from 1 to 8 channels per stream. Channels are  
only grouped in FLAC to take advantage of interchannel correlation  
and to define common channel assignments (like stereo L/R, 5.1  
surround, et cetera). When encoding a large number of independent  
channels it is expected that they are coded separately and if  
required, multiplexed together in a suitable container like Ogg or  
Matroska."


Good info.

 But this is a bit typical as an “answer”: HOW would you code many  
channels separately in Flac (available sw to use etc.), and how to  
multiplex into a “suitable container”?


 Ok, no problem for anybody else... I am supposed to see! 

 (https://xiph.org/flac/ogg_mapping.html

 They don’t explain this at all, unless you take the faq answer as  
a definition.)


 Best,

 Stefan

 P.S.: They just have to extend (too) old definitions.

 As Fons says, the limit is a limit (paraphrasing Theresa May) -  
and not justified from a current perspective.


A few years ago I hacked the opusenc command line tool to encode  
channels without inter-channel correlation, because the correlations  
in the Opus codec were programmed for specific multi-channel  
schemes, limited to 8 channels. The "joint stereo" MP3 mode is an  
example of such a scheme, where channels are internally converted  
before encoding (see:  
https://en.wikipedia.org/wiki/Joint_%28audio_engineering%29).The  
Opus codec now have an Ambisonics mode with new inter-channel  
correlation schemes. I don't know if the fine Xiph developers can  
"just" extend the definitions of FLAC, if a special Ambisonics mode  
would be required, and to what extent the 8-channel limit (as a  
limit) is a political issue. For the multiplexing of multiple  
streams in a suitable container, it is left as an exercise. And if  
the issue is just disk space, file compression softwares (zip, gzip,  
etc) are still valid tools.




 Marc


- - - - -

 1. I believe that the opus encoders/decoders have always supported  
more than 8 channels.


2. The next question is what ogg channel mapping and consequently  
real-world browsers allow...


But in some sense the hack you did is known. (More complicated is  
maybe to make it work...)


3. If they already plan to issue some ogg ambisonics standard (using  
ogg opus of course) since at least 2016: You also need an associated  
mastering standard, which would not change or compress any audio data.  
Correct?


So what is “political” about extending the channel count of FLAC?

Compromise proposal:

4. So let’s maybe use .wav or .caf for the “mastering format”.  
Microsoft and Apple already allow more than 8 channels... 樂


Best,

Stefan

P.S.: “Joint stereo” you could classify as parametric coding.
-- next part --
An HTML attachment was scrubbed...
URL: 

___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.

Re: [Sursound] octofile release

[Marc Lavallée]

Le 2018-07-29 à 03:08 PM, Stefan Schreiber a écrit :

Citando Marc Lavallée :

Excerpt from https://xiph.org/flac/faq.html#general__channels :
 "FLAC supports from 1 to 8 channels per stream. Channels are only 
grouped in FLAC to take advantage of interchannel correlation and to 
define common channel assignments (like stereo L/R, 5.1 surround, et 
cetera). When encoding a large number of independent channels it is 
expected that they are coded separately and if required, multiplexed 
together in a suitable container like Ogg or Matroska."


Good info.
But this is a bit typical as an “answer”: HOW would you code many 
channels separately in Flac (available sw to use etc.), and how to 
multiplex into a “suitable container”?

Ok, no problem for anybody else... I am supposed to see! 
(https://xiph.org/flac/ogg_mapping.html
They don’t explain this at all, unless you take the faq answer as a 
definition.)

Best,
Stefan
P.S.: They just have to extend (too) old definitions.
As Fons says, the limit is a limit (paraphrasing Theresa May) - and 
not justified from a current perspective.


A few years ago I hacked the opusenc command line tool to encode 
channels without inter-channel correlation, because the correlations in 
the Opus codec were programmed for specific multi-channel schemes, 
limited to 8 channels. The "joint stereo" MP3 mode is an example of such 
a scheme, where channels are internally converted before encoding (see: 
https://en.wikipedia.org/wiki/Joint_%28audio_engineering%29).The Opus 
codec now have an Ambisonics mode with new inter-channel correlation 
schemes. I don't know if the fine Xiph developers can "just" extend the 
definitions of FLAC, if a special Ambisonics mode would be required, and 
to what extent the 8-channel limit (as a limit) is a political issue. 
For the multiplexing of multiple streams in a suitable container, it is 
left as an exercise. And if the issue is just disk space, file 
compression softwares (zip, gzip, etc) are still valid tools.


Marc




 Le 2018-07-29 à 02:20 PM, Fons Adriaensen a écrit :


On Sun, Jul 29, 2018 at 01:46:04PM -0400, Marc Lavallée wrote:


Easy to compile, support for FLAC, GPL3 license...


FLAC is supported for input only. The output is 9 channels,

 and FLAC can't handle that. Never understood why they put

 in that silly limit...



 Ciao,


___

 Sursound mailing list

surso...@music.vt.eduhttps://mail.music.vt.edu/mailman/listinfo/sursound[1] 
- unsubscribe here, edit account or options, view archives and so on.




-- next part --
An HTML attachment was scrubbed...
URL: 

___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.

Re: [Sursound] octofile release

[Stefan Schreiber]

Citando Marc Lavallée :


Excerpt from https://xiph.org/flac/faq.html#general__channels :



 "FLAC supports from 1 to 8 channels per stream. Channels are only  
grouped in FLAC to take advantage of interchannel correlation and to  
define common channel assignments (like stereo L/R, 5.1 surround, et  
cetera). When encoding a large number of independent channels it is  
expected that they are coded separately and if required, multiplexed  
together in a suitable container like Ogg or Matroska."


Good info.

But this is a bit typical as an “answer”: HOW would you code many  
channels separately in Flac (available sw to use etc.), and how to  
multiplex into a “suitable container”?


Ok, no problem for anybody else... I am supposed to see! 

(https://xiph.org/flac/ogg_mapping.html

They don’t explain this at all, unless you take the faq answer as a  
definition.)


Best,

Stefan

P.S.: They just have to extend (too) old definitions. 

As Fons says, the limit is a limit (paraphrasing Theresa May) - and  
not justified from a current perspective. 



Marc





 Le 2018-07-29 à 02:20 PM, Fons Adriaensen a écrit :


On Sun, Jul 29, 2018 at 01:46:04PM -0400, Marc Lavallée wrote:


Easy to compile, support for FLAC, GPL3 license...


FLAC is supported for input only. The output is 9 channels,

 and FLAC can't handle that. Never understood why they put

 in that silly limit...



 Ciao,


___

 Sursound mailing list

  
surso...@music.vt.eduhttps://mail.music.vt.edu/mailman/listinfo/sursound[1]  
- unsubscribe here, edit account or options, view archives and so on.




Ligações:
-
[1]
-- next part --
An HTML attachment was scrubbed...
URL: 

___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.

Re: [Sursound] octofile release

[Marc Lavallée]

Excerpt from https://xiph.org/flac/faq.html#general__channels :

"FLAC supports from 1 to 8 channels per stream. Channels are only 
grouped in FLAC to take advantage of interchannel correlation and to 
define common channel assignments (like stereo L/R, 5.1 surround, et 
cetera). When encoding a large number of independent channels it is 
expected that they are coded separately and if required, multiplexed 
together in a suitable container like Ogg or Matroska."


Marc


Le 2018-07-29 à 02:20 PM, Fons Adriaensen a écrit :

On Sun, Jul 29, 2018 at 01:46:04PM -0400, Marc Lavallée wrote:


Easy to compile, support for FLAC, GPL3 license...

FLAC is supported for input only. The output is 9 channels,
and FLAC can't handle that. Never understood why they put
in that silly limit...

Ciao,



___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.

Re: [Sursound] octofile release

[Fons Adriaensen]

On Sun, Jul 29, 2018 at 01:46:04PM -0400, Marc Lavallée wrote:

> Easy to compile, support for FLAC, GPL3 license...

FLAC is supported for input only. The output is 9 channels,
and FLAC can't handle that. Never understood why they put
in that silly limit...

Ciao,

-- 
FA


___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.

Re: [Sursound] octofile release

[Marc Lavallée]

Easy to compile, support for FLAC, GPL3 license...

Thanks!

Marc


Le 2018-07-29 à 11:02 AM, Fons Adriaensen a écrit :

Hello all,

Source code for octofile version 0.3.0 (linux) is now available at



Octofile is the A to B format converter for Core Sound's Octomic.

A-format input can be 1,2,4 or 8 audio file(s) with resp. 8,4,2
or 1 channel(s) each, 44.1, 48 or 96 kHz.

Default output is a 2nd order Ambix file (CAF, SN3D, ACN) but the
legacy .amb or non-standard formats (e.g. Ambix in a .wav file)
are possible as well.

You will require a calibration file from Core Sound of course.

Ciao,



___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.

[Sursound] octofile release

[Fons Adriaensen]

Hello all,

Source code for octofile version 0.3.0 (linux) is now available at

 

Octofile is the A to B format converter for Core Sound's Octomic.

A-format input can be 1,2,4 or 8 audio file(s) with resp. 8,4,2
or 1 channel(s) each, 44.1, 48 or 96 kHz.

Default output is a 2nd order Ambix file (CAF, SN3D, ACN) but the
legacy .amb or non-standard formats (e.g. Ambix in a .wav file)
are possible as well.

You will require a calibration file from Core Sound of course.

Ciao,

-- 
FA


___
Sursound mailing list
Sursound@music.vt.edu
https://mail.music.vt.edu/mailman/listinfo/sursound - unsubscribe here, edit 
account or options, view archives and so on.