This presentation, created for the 2007 Audio Engineering Society UK Conference, describes how to distribute superior Ambisonic surround-sound audio content in a way that maximises its accessibility by requiring only standard surround replay equipment. It's a longer version of Part 4 of my Ionian University presentations.
5. Ambisonics
• Invented by Michael Gerzon
and associates in the 1970s.
• Is not based on ‘one channel/one speaker’:
6. Ambisonics
• Invented by Michael Gerzon
and associates in the 1970s.
• Is not based on ‘one channel/one speaker’:
• Instead, soundfield is captured or created,
transmitted and rendered at the listening
end for the speaker array available.
7. Ambisonics
• Invented by Michael Gerzon
and associates in the 1970s.
• Is not based on ‘one channel/one speaker’:
• Instead, soundfield is captured or created,
transmitted and rendered at the listening
end for the speaker array available.
• Analogous to a Postscript file/printer.
9. Ambisonics
• Two methods of transmission:
• B-Format (the studio
format):
sum & difference channels.
10. Ambisonics
• Two methods of transmission:
• B-Format (the studio
format):
sum & difference channels.
• UHJ stereo-compatible
matrix:
(typically 2-channel UHJ).
14. Ambisonic Mixing
• The vast majority of
popular music is mixed
from multitrack.
• Ambisonics is particularly
good at this: for example,
it’s isotropic.
15. Ambisonic Mixing
• The vast majority of
popular music is mixed
from multitrack.
• Ambisonics is particularly
good at this: for example,
it’s isotropic.
• Main interest of
‘Ambisonic Mixing Group’.
17. Discounted options
for distributing Ambisonic recordings
• MLP can carry B-Format and offer flags
for DVD-Audio, HD discs
18. Discounted options
for distributing Ambisonic recordings
• MLP can carry B-Format and offer flags
for DVD-Audio, HD discs
• …sorry, not very many people can play them.
19. Discounted options
for distributing Ambisonic recordings
• MLP can carry B-Format and offer flags
for DVD-Audio, HD discs
• …sorry, not very many people can play them.
• Downloadable Ambisonic files
and software players
20. Discounted options
for distributing Ambisonic recordings
• MLP can carry B-Format and offer flags
for DVD-Audio, HD discs
• …sorry, not very many people can play them.
• Downloadable Ambisonic files
and software players
• …here or on the way – but not yet widespread.
22. Goals
for distributing Ambisonic recordings
• Method must be available now, today.
23. Goals
for distributing Ambisonic recordings
• Method must be available now, today.
• Everyone (with a surround system)
can listen.
24. Goals
for distributing Ambisonic recordings
• Method must be available now, today.
• Everyone (with a surround system)
can listen.
• Easy and cheap to originate content.
26. The Answer…
Decoding in the studio
• Decode Ambisonics to speaker feeds.
27. The Answer…
Decoding in the studio
• Decode Ambisonics to speaker feeds.
• Place the speaker feeds signals on a
conventional multichannel medium
(eg DVD).
28. The Answer…
Decoding in the studio
• Decode Ambisonics to speaker feeds.
• Place the speaker feeds signals on a
conventional multichannel medium
(eg DVD).
• This is called…
29. G-Format
Decoding Ambisonics in the studio
G
…to 5.1-compatible multichannel surround.
31. G-Format
Advantages
• Anyone can play it (if you choose popular
encoding and distribution formats).
32. G-Format
Advantages
• Anyone can play it (if you choose popular
encoding and distribution formats).
• Producer controls the decoding process.
33. G-Format
Advantages
• Anyone can play it (if you choose popular
encoding and distribution formats).
• Producer controls the decoding process.
• High quality decoder can be used,
correctly configured.
35. G-Format
Challenges
• Lose the ability to render for your replay
array (a primary Ambisonic benefit).
36. G-Format
Challenges
• Lose the ability to render for your replay
array (a primary Ambisonic benefit).
• With some caveats, decode can be reversible.
37. G-Format
Challenges
• Lose the ability to render for your replay
array (a primary Ambisonic benefit).
• With some caveats, decode can be reversible.
• What array do you decode to?
38. G-Format
Challenges
• Lose the ability to render for your replay
array (a primary Ambisonic benefit).
• With some caveats, decode can be reversible.
• What array do you decode to?
• Decoder target array and listener array
may be different.
39. G-Format
Challenges
• Lose the ability to render for your replay
array (a primary Ambisonic benefit).
• With some caveats, decode can be reversible.
• What array do you decode to?
• Decoder target array and listener array
may be different.
• Possible risk of audio and/or localisation
degradation if lossy compression used.
40. G-Format
What decode array?
N
•
C
ITU? L R
30º
100–120º
Ls Rs
41. G-Format
What decode array?
N
•
C
ITU? L R
• Decoding Ambisonics 30º
to ITU 5.1 is
technically possible.
100–120º
Ls Rs
42. G-Format
What decode array?
N
•
C
ITU? L R
• Living rooms 30º
aren’t good 100–120º
for ITU.
Ls Rs
43. G-Format
What decode array?
N
•
C
ITU? L R
• Living rooms 30º
aren’t good 100–120º
for ITU.
• We have Ls Rs
rectangular
living rooms.
44. G-Format
Decode array
N
L R
• Rectangular Array.
Ls Rs
45. G-Format
Decode array
N
L R
• Rectangular Array.
• A standard
Ambisonic
configuration.
Ls Rs
46. G-Format
Decode array
• But narrow
N
or wide?
L R
Ls Rs
47. G-Format
Decode array
• But narrow
N
or wide?
L R
• 4-speaker
Ambisonics
supports side
ratios
between
1:2 & 2:1 Ls Rs
48. G-Format
Decode array
• We don’t know.
N
L R
Ls Rs
49. G-Format
Decode array
• We don’t know.
N
• A square layout L R
minimises
likely errors.
Ls Rs
50. G-Format
Decode array
• We don’t know.
N
• A square layout L R
minimises
likely errors.
• Nimbus also
found square
best. Ls Rs
53. G-Format
Decoder array and listener array may be different.
• Fronts wider than ITU; N
• Rears are narrower. L R
Ls Rs
54. G-Format
Decoder array and listener array may be different.
• Fronts wider than ITU; N
• Rears are narrower. L R
• But listeners may
well not have ITU!
Ls Rs
55. G-Format
Decoder array and listener array may be different.
• Fronts wider than ITU; N
• Rears are narrower. L R
• But listeners may
well not have ITU!
• Ambisonics is robust.
Ls Rs
56. G-Format
Decoder array and listener array may be different.
• Fronts wider than ITU; N
• Rears are narrower. L R
• But listeners may
well not have ITU!
• Ambisonics is robust.
• CF is not essential
(LFE not needed Ls Rs
at all!).
58. G-Format Sources
B-Format or 2-ch UHJ?
• Very little B-format originally
(No multitrack mastering capability)
59. G-Format Sources
B-Format or 2-ch UHJ?
• Very little B-format originally
(No multitrack mastering capability)
• This is now changing.
60. G-Format Sources
B-Format or 2-ch UHJ?
• Very little B-format originally
(No multitrack mastering capability)
• This is now changing.
• UHJ recordings are very common.
61. G-Format Sources
B-Format or 2-ch UHJ?
• Very little B-format originally
(No multitrack mastering capability)
• This is now changing.
• UHJ recordings are very common.
• How would they stand up against
modern multichannel?
62. G-Format Sources
B-Format or 2-ch UHJ?
• Very little B-format originally
(No multitrack mastering capability)
• This is now changing.
• UHJ recordings are very common.
• How would they stand up against
modern multichannel?
• Nimbus showed SFM recordings work.
What about Ambisonic mixes?
65. What decoder?
• Decode needed to be done in software.
• Several decoders available on-line
(see www.ambisonic.net → tools).
66. What decoder?
• Decode needed to be done in software.
• Several decoders available on-line
(see www.ambisonic.net → tools).
• We chose Ambisonic.exe written by
Rhonda Wilson at Meridian Audio.
67. What decoder?
• Decode needed to be done in software.
• Several decoders available on-line
(see www.ambisonic.net → tools).
• We chose Ambisonic.exe written by
Rhonda Wilson at Meridian Audio.
• Originally created for Nimbus.
68. What decoder?
• Decode needed to be done in software.
• Several decoders available on-line
(see www.ambisonic.net → tools).
• We chose Ambisonic.exe written by
Rhonda Wilson at Meridian Audio.
• Originally created for Nimbus.
• UHJ and B-Format capability.
69. What decoder?
• Decode needed to be done in software.
• Several decoders available on-line
(see www.ambisonic.net → tools).
• We chose Ambisonic.exe written by
Rhonda Wilson at Meridian Audio.
• Originally created for Nimbus.
• UHJ and B-Format capability.
• Decodes to a rectangle (2:1 – 1:2).
72. Release coding & media
Requirements
• Available now.
• Not a future system.
73. Release coding & media
Requirements
• Available now.
• Not a future system.
• Thus not surround downloads
(ie must be a ‘packaged medium’).
74. Release coding & media
Requirements
• Available now.
• Not a future system.
• Thus not surround downloads
(ie must be a ‘packaged medium’).
• Widely supported.
75. Release coding & media
Requirements
• Available now.
• Not a future system.
• Thus not surround downloads
(ie must be a ‘packaged medium’).
• Widely supported.
• Must be popular and widespread.
76. Release coding & media
Requirements
• Available now.
• Not a future system.
• Thus not surround downloads
(ie must be a ‘packaged medium’).
• Widely supported.
• Must be popular and widespread.
• Thus not SACD, DVD-A.
87. Release medium
Two choices: DVD & CD
• DVD-Video:
• Authoring application required as well as
surround encoder.
88. Release medium
Two choices: DVD & CD
• DVD-Video:
• Authoring application required as well as
surround encoder.
• Images required (even if rudimentary).
89. Release medium
Two choices: DVD & CD
• DVD-Video:
• Authoring application required as well as
surround encoder.
• Images required (even if rudimentary).
• Time required to build menu structure
etc.
91. Release medium
Two choices: DVD & CD
• CD-DTS
• Encoder costs $99
(Surcode).
92. Release medium
Two choices: DVD & CD
• CD-DTS
• Encoder costs $99
(Surcode).
• Very quick to author.
93. Release medium
Two choices: DVD & CD
• CD-DTS
• Encoder costs $99
(Surcode).
• Very quick to author.
• Burn with simple PC CD-burning
application (eg EAC, Nero).
94. Release medium
Two choices: DVD & CD
• CD-DTS
• Encoder costs $99
(Surcode).
• Very quick to author.
• Burn with simple PC CD-burning
application (eg EAC, Nero).
• Very low production cost
(CD-R or pressing).
97. Will G-Format work?
• Nimbus showed
UHJ single-point
mic recordings
worked.
• Early tests
suggested that
UHJ mixes didn’t.
98. Will G-Format work?
• Nimbus showed
UHJ single-point
mic recordings
worked.
• Early tests
suggested that
UHJ mixes didn’t.
• Luckily, early tests
were wrong.
101. First test disc
• Started with the first Ambisonic
mixes ever made…
• …KPM Music Library tracks
mixed to UHJ for the Cunard hi-fi
show in 1981.
102. First test disc
• Started with the first Ambisonic
mixes ever made…
• …KPM Music Library tracks
mixed to UHJ for the Cunard hi-fi
show in 1981.
• Sourced initially from cassette.
103. First test disc
• Started with the first Ambisonic
mixes ever made…
• …KPM Music Library tracks
mixed to UHJ for the Cunard hi-fi
show in 1981.
• Sourced initially from cassette.
• Success!
105. Initial results
Planar B-Format
• B-Format test recordings from
www.ambisonicbootlegs.net.
106. Initial results
Planar B-Format
• B-Format test recordings from
www.ambisonicbootlegs.net.
• Mixture of SFM and digital mixes.
107. Initial results
Planar B-Format
• B-Format test recordings from
www.ambisonicbootlegs.net.
• Mixture of SFM and digital mixes.
• Extremely impressive results.
(UHJ and B-Format sourced demos
played in the Council Room)
108. Full-length discs
UHJ
• Surprise Surprise
• Stereotomy
• KPM Library discs
• Collins Classics CDs
• Nimbus, Music from York
…and more.
109. Full-length discs
UHJ
• Surprise Surprise
• Stereotomy
• KPM Library discs
• Collins Classics CDs
• Nimbus, Music from York
…and more.
110. Full-length discs
UHJ
• Surprise Surprise
• Stereotomy
• KPM Library discs
• Collins Classics CDs
• Nimbus, Music from York
…and more.
111. Full-length discs
B-Format
• Nimbus NI5763:
Bridge, Elgar, Holst:
Concertos for Cello & Orchestra
Raphael Walfisch, cello;
Royal Liverpool Philharmonic Orchestra
conducted by Richard Dickins.
112. Full-length discs
B-Format
• Nimbus NI5763:
Bridge, Elgar, Holst:
Concertos for Cello & Orchestra
Raphael Walfisch, cello;
Royal Liverpool Philharmonic Orchestra
conducted by Richard Dickins.
• Halliday mic array
at Philharmonic Hall,
planar B-Format.
113. Full-length discs
B-Format
• Nimbus NI5763:
Bridge, Elgar, Holst:
Concertos for Cello & Orchestra
Raphael Walfisch, cello;
Royal Liverpool Philharmonic Orchestra
conducted by Richard Dickins.
• Halliday mic array
at Philharmonic Hall,
planar B-Format.
115. Tweaks
Adjusting the workflow
• Noise bursts between
tracks an issue.
116. Tweaks
Adjusting the workflow
• Noise bursts between
tracks an issue.
• Led to assembly of
complete disc master as a
single CD-DTS file.
117. Tweaks
Adjusting the workflow
• Noise bursts between
tracks an issue.
• Led to assembly of
complete disc master as a
single CD-DTS file.
• Use of .CUE files to tell
burning application the
TNO locations.
118. Final workflow
• Assemble tracks as multichannel or
individual WAV files. (Disentangle .AMB)
• Concatenate with digital silence between
tracks to create album file(s).
• Create .CUE file and edit as required.
• Decode to speaker feeds.
• Encode to CD-DTS.
• Burn disc from CUE file.
119. Future plans
• Create new B-Format mixes…
• Decode and release them!
•
Mastering system… VST plugins/Nuendo
• Multitrack material welcome!
Intro slide. On screen as early as possible.\nLeft click adds points/advances slide. Right-click removes points/goes to previous slide.\n
Mouse click to add each new point on a slide, as well as advancing slides. First click here brings up Gerzon pic; click immediately again to bring up 1st point. Then, 3 points skipping over Ambisonic principles. Go as fast as possible.\n
Mouse click to add each new point on a slide, as well as advancing slides. First click here brings up Gerzon pic; click immediately again to bring up 1st point. Then, 3 points skipping over Ambisonic principles. Go as fast as possible.\n
Mouse click to add each new point on a slide, as well as advancing slides. First click here brings up Gerzon pic; click immediately again to bring up 1st point. Then, 3 points skipping over Ambisonic principles. Go as fast as possible.\n
Mouse click to add each new point on a slide, as well as advancing slides. First click here brings up Gerzon pic; click immediately again to bring up 1st point. Then, 3 points skipping over Ambisonic principles. Go as fast as possible.\n
Mouse click to add each new point on a slide, as well as advancing slides. First click here brings up Gerzon pic; click immediately again to bring up 1st point. Then, 3 points skipping over Ambisonic principles. Go as fast as possible.\n
Contrasts B-Format and UHJ. Three points (one main which comes up with slide). Go through as far as possible.\n
Contrasts B-Format and UHJ. Three points (one main which comes up with slide). Go through as far as possible.\n
Contrasts B-Format and UHJ. Three points (one main which comes up with slide). Go through as far as possible.\n
Conventional mic-based view of Ambisonics. Just the one point, on long enough to say it. Slide comes up complete.\n
“Why is Ambisonic Mixing important?” Say this and then click to bring up point 1: Because most music is done that way. Isotropic = gives equal precedent from sounds coming from all directions - important for artificial soundfields. ‘AMG’ is us of course. 3 points. An old picture of RE.\n
“Why is Ambisonic Mixing important?” Say this and then click to bring up point 1: Because most music is done that way. Isotropic = gives equal precedent from sounds coming from all directions - important for artificial soundfields. ‘AMG’ is us of course. 3 points. An old picture of RE.\n
“Why is Ambisonic Mixing important?” Say this and then click to bring up point 1: Because most music is done that way. Isotropic = gives equal precedent from sounds coming from all directions - important for artificial soundfields. ‘AMG’ is us of course. 3 points. An old picture of RE.\n
Other methods of Ambisonic distro: not appropriate for one reason or another. 4 points (two main). MLP = Meridian Lossless Packing.\n
Other methods of Ambisonic distro: not appropriate for one reason or another. 4 points (two main). MLP = Meridian Lossless Packing.\n
Other methods of Ambisonic distro: not appropriate for one reason or another. 4 points (two main). MLP = Meridian Lossless Packing.\n
Other methods of Ambisonic distro: not appropriate for one reason or another. 4 points (two main). MLP = Meridian Lossless Packing.\n
Here’s what we were looking for as an alternative. Three points\n
Here’s what we were looking for as an alternative. Three points\n
Here’s what we were looking for as an alternative. Three points\n
The solution we chose. Named on next slide. 3 points. When final point comes up, say “We call this...” and advance slide...\n
The solution we chose. Named on next slide. 3 points. When final point comes up, say “We call this...” and advance slide...\n
The solution we chose. Named on next slide. 3 points. When final point comes up, say “We call this...” and advance slide...\n
Say, “G FORMAT”\nJust the one item\n
So here are the advantages of this approach... 3 points\n
So here are the advantages of this approach... 3 points\n
So here are the advantages of this approach... 3 points\n
...and here are the drawbacks. 5 points (first has a sub): you may be able to get B-Format back from G, then decode Ambisonically and get this benefit back. Needs reversible shelf filters. \n
...and here are the drawbacks. 5 points (first has a sub): you may be able to get B-Format back from G, then decode Ambisonically and get this benefit back. Needs reversible shelf filters. \n
...and here are the drawbacks. 5 points (first has a sub): you may be able to get B-Format back from G, then decode Ambisonically and get this benefit back. Needs reversible shelf filters. \n
...and here are the drawbacks. 5 points (first has a sub): you may be able to get B-Format back from G, then decode Ambisonically and get this benefit back. Needs reversible shelf filters. \n
...and here are the drawbacks. 5 points (first has a sub): you may be able to get B-Format back from G, then decode Ambisonically and get this benefit back. Needs reversible shelf filters. \n
This section addresses the challenges one by one. First, the decode array: why we chose four-square. One point plus sub\n
This section addresses the challenges one by one. First, the decode array: why we chose four-square. One point plus sub\n
Why ITU was not what we chose. 3 points. “ITU” stays up from last slide.\n
Why ITU was not what we chose. 3 points. “ITU” stays up from last slide.\n
Why ITU was not what we chose. 3 points. “ITU” stays up from last slide.\n
Rectangular array - two points\n
Rectangular array - two points\n
Aspect rations. Initial question answered on next slide. 2 points\n
Aspect rations. Initial question answered on next slide. 2 points\n
Aspect rations. Initial question answered on next slide. 2 points\n
We don’t know whether the listening room will be narrow or wide: a square target minimises the likely error.\n
We don’t know whether the listening room will be narrow or wide: a square target minimises the likely error.\n
We don’t know whether the listening room will be narrow or wide: a square target minimises the likely error.\n
The difference between target and listener arrays. 5 points!\n
The difference between target and listener arrays. 5 points!\n
The difference between target and listener arrays. 5 points!\n
The difference between target and listener arrays. 5 points!\n
The difference between target and listener arrays. 5 points!\n
Source material - 5 points. Question of effectiveness of UHJ decodes.\n
Source material - 5 points. Question of effectiveness of UHJ decodes.\n
Source material - 5 points. Question of effectiveness of UHJ decodes.\n
Source material - 5 points. Question of effectiveness of UHJ decodes.\n
Source material - 5 points. Question of effectiveness of UHJ decodes.\n
Decoder choice. 6 points\n
Decoder choice. 6 points\n
Decoder choice. 6 points\n
Decoder choice. 6 points\n
Decoder choice. 6 points\n
Decoder choice. 6 points\n
Criteria for a release medium and its coding. 6 points.\n
Criteria for a release medium and its coding. 6 points.\n
Criteria for a release medium and its coding. 6 points.\n
Criteria for a release medium and its coding. 6 points.\n
Criteria for a release medium and its coding. 6 points.\n
Criteria for a release medium and its coding. 6 points.\n
Dolby Digital - considerations (4). Discuss that there is some doubt about Dolby’s compatibility with Ambisonics - coupling may impact phase relationships.\n
Dolby Digital - considerations (4). Discuss that there is some doubt about Dolby’s compatibility with Ambisonics - coupling may impact phase relationships.\n
Dolby Digital - considerations (4). Discuss that there is some doubt about Dolby’s compatibility with Ambisonics - coupling may impact phase relationships.\n
Dolby Digital - considerations (4). Discuss that there is some doubt about Dolby’s compatibility with Ambisonics - coupling may impact phase relationships.\n
DTS - pro points (4)\n
DTS - pro points (4)\n
DTS - pro points (4)\n
DTS - pro points (4)\n
This was our choice. Again, note that Nimbus chose DTS for the DVD-Video zone of their G-Format discs.\n
Release medium choices #1: DVD-Video - 4 points. \n
Release medium choices #1: DVD-Video - 4 points. \n
Release medium choices #1: DVD-Video - 4 points. \n
Release medium choices #1: DVD-Video - 4 points. \n
CD-DTS considerations - 5 points\n
CD-DTS considerations - 5 points\n
CD-DTS considerations - 5 points\n
CD-DTS considerations - 5 points\n
CD-DTS considerations - 5 points\n
3 points - refer to early experiments where we had a test version of Stereotomy made for us.\n
3 points - refer to early experiments where we had a test version of Stereotomy made for us.\n
3 points - refer to early experiments where we had a test version of Stereotomy made for us.\n
4 points. Refer to Boots Audio project. Started with UHJ.\n
4 points. Refer to Boots Audio project. Started with UHJ.\n
4 points. Refer to Boots Audio project. Started with UHJ.\n
4 points. Refer to Boots Audio project. Started with UHJ.\n
Then some B Format sources - even better.\n
Then some B Format sources - even better.\n
Then some B Format sources - even better.\n
Then we processed some complete discs and refined the workflow. Details are in the paper. All points come up at once; click to display Surprise album cover and again for Stereotomy\n
Then we processed some complete discs and refined the workflow. Details are in the paper. All points come up at once; click to display Surprise album cover and again for Stereotomy\n
Then we were very kindly offered an original B-Format Nimbus recording to try. Thanks to Crac Downes at Nimbus. You can hear some of this in the Council Room demos. 2 points. \n
Then we were very kindly offered an original B-Format Nimbus recording to try. Thanks to Crac Downes at Nimbus. You can hear some of this in the Council Room demos. 2 points. \n
Then we were very kindly offered an original B-Format Nimbus recording to try. Thanks to Crac Downes at Nimbus. You can hear some of this in the Council Room demos. 2 points. \n
Further development of workflow to allow processing of one (series of) files. 3 points. Noise bursts due to dts decoder de-muting.\n
Further development of workflow to allow processing of one (series of) files. 3 points. Noise bursts due to dts decoder de-muting.\n
Further development of workflow to allow processing of one (series of) files. 3 points. Noise bursts due to dts decoder de-muting.\n
6 points. Covered in much more detail in the paper. “Disentangle” means to turn multichannel WAV files into individual mono files as our apps don’t work with the former.\n