Header image  
by Ralph Glasgal
 
line decor
Home Tutorials Tech
Papers
Kudos and
Pictures
Demos Bio Free Ambio
book
Glossary The Home
Concert Hall
PC/Mobile
Applications
Rec Engineers
Corner
FAQ/Forum Links Contact us
line decor

Audio Engineering Society
Convention Paper

Transforming Ambiophonic + Ambisonic 3D Surround Sound to & from ITU 5.1/6.1
Robert E. (Robin) Miller III ©20031

Page 6 -- RECONSTITUTING FULL PERIPHONIC 3D

The year is not 1492; the world of sound is not flat!  To implement "periphonic" full sphere 3D surround for an uncompromised experience, the author has determined experimentally [2] a minimum requirement of 10 speakers (plus subwoofer).  From the six full-range media channels (plus optional LFE) can be derived feeds for any number of speakers, but there is diminishing return above 10, illustrated in Fig.10.

In order for consumers to recover PerAmbio periphonic 3D from the dual-format recording, a žsmart decoderÓ with DSP is required, controlled by one of six (or more) reconstitution matrices, selected by detecting a flag in metadata.  Then an ambience speaker feed matrix and at least 8 speakers if not 12 or 24 are required, as shown for 8 in Table 13, in addition to the Ambiophonic pair FL,FR, plus subwoofer(s).


Table 13. Matrix to feed an 8-speaker Ambisonic array, a regular cube tilted forward 45° for PerAmbio 6.1.10+ (Fig.10).  W =1.000 because 0.707 (-3dB) normal at this stage was applied in transformation {S}.

PerAmbio 6.1.10+ Speaker Layout (3D)

The minimum PerAmbio layout is shown in Fig. 10.  It is described as a main Ambiophonic pair (FL, FR) plus a regular cubic Ambisonic array of eight speakers tilted forward 45° [2].  Additional speakers may be placed isotropically so the ambience array becomes a cuboctahedron (12 speakers) or a practical maximum of 24 speakers in a dodecahedral array.


Fig.10  PerAmbio 6.1.10+ with 10 speakers plus subwoofer(s) - white at ear level, black on floor, gray at ceiling. Ambience can grow from 8 to 12 speakers (cuboctahedron), or to 24 (dodecahedron).

PerAmbio 6.1.10+ delivery media (3D)

Whether using the minimum of 10 or a greater array of 14 or 26 speakers, + subwoofer(s), the approach is named here "PerAmbio 6.1.10+" - see žMultichannel NumberingÓ below.  Music can be delivered to users today using DVD-A, SACD, or DTS-ES Discrete.  Current DVD-V can also transport PerAmbio using DTS 6.1 ES Discrete, but not using Dolby Digital 5.1 or DD-EX because it derives SC from a matrix.

PerAmbio 3D surround cannot be delivered to television audiences with current implementations for terrestrial Advanced Television (ATV) broadcast or cablecast, nor with DVD-V (except using DTS-ES 6.1 Discrete) [17].  These standards would need to be updated for 3D, or implemented e.g. using MPEG4 AAC for ATV [18].  The DVD-V standard provides for 8 channels that also have yet to be implemented.  However, while DD/AC-3 in currently implemented ATV and DVD-V standards is incapable of PerAmbio 3D, DD/AC-3 can transport another Ambiophonic-based 2D surround reproduction technique, PanAmbio 4.1, which is compatible with ITU 5.1 and offers tools for 5.1 production, as explored in prior papers [2,4,7].

Multichannel Numbering in the Form r.lfe.s

Time was the number of production channels equaled the number of media channels, which in turn equaled the number of speakers.  In the new order, the number of production channels, number of media channels, and number of speakers can all be different!

Consumer marketers have blurred the standard designations 5.1, 6.1 etc. to mean žnumber of speakers,Ó not discrete production or media channels.  Reputable manufacturers avoid a numbers game: Dolby officially does not designate DD-EX (matrix surround center) as 6.1, as it is really still 5.1 channels.  DTS-ES 6.1 Discrete is a true 6.1 channel system. 

To clarify that fewer media channels "r" are required than speakers "s" (when their feeds can be derived during replay), the author proposes [2] use of MCN - multichannel numbering, in the form "r.lfe.s" - which simply appends to the usual number of media channels, r (mnemonic žradioedÓ), a second decimal point and number, s, to indicate unique speaker feeds.

Examples are: 5.1.5 for ITU standard 5.1; 5.1.6 for Dolby EX (SC matrixed from LR, RS); 6.1.6 for DTS-ES 6.1 Discrete, etc.  PerAmbio 3D is 6.0.10+ varying with the user's number of ambience speakers and feeds (the recording is Ambisonically encoded for any number).  Subwoofer(s) are always implied depending on the user's installation.

BEYOND 5.1: PERIPHONIC 3D SURROUND?

With the exception of IMAX, žmultichannel surround soundÓ today means surround in the horizontal plane.  Spawned by cinema, only left and right boundaries of the picture frame are extended by the sound žimageÓ to full circle; vertical sound is contributed only by reverberation of the theater (significant if large) or home theater (ranging from not much to awful if untreated).  This definition of žsurroundÓ was predicated on the oversimplification that humans have but two ears on a horizontal axis.

However, human ears also have pinna.  These personalized appendages are direction finders for confirming fine horizontal localization to about 1°, but also cueing height above and below the plane of the ears.  So future music recording, Advanced Television, and cinema sound need to expand the definition of žsurroundÓ to include height.  Indeed, IMAXŪs five-story screen hides a single height speaker near its top.  One need only experience full sphere surround with height, as the author demonstrates daily (Fig. 11), to recognize that its contribution is as important to the ear as a jump in picture resolution of HDTV is to the eye.

Why Height?

When we experience live events, our eyes+brain system resolves an image with žpixelsÓ measuring half a minute of arc, exceeding high definition television at normal viewing distances.  And we are immersed in a bubble of pulsing air, where, at for example a sporting event, we expect to hear the crowd elevated and the action below.  Or the inverse - an air show above and the crowd below, or a lofty pipe organ with choir below.  In a church, we would surely notice if organ and choir were squashed together at the horizon.  Or if at the opera, suddenly the ceiling and floor were on the same plane, so that all the reflections weŪve come to expect to be strongly and spherically directional would flatten, erasing much of the acoustic signature of the hall, along with our envelopment within and enjoyment of it.  We expect in real life our ears+brain system to resolve a sonic žimageÓ with acuity of one degree horizontal and several degrees vertical - due to changes in comb filtering by the pinna with varying elevation.  In real life, we would be astonished, were 3D sound suddenly absent!

Conversely, since we can not yet expect this level of realism from our home theater, we are just as astonished when we hear our first reproduction of full periphony - 3D sound that blows away the walls of the listening room, coming at us from everywhere in the sphere - not just the circle - wherein we are the psychoacoustic center.  If not direct sources above and below (front as well as back), directional room reflections affect our perception of žnaturalÓ tonal color of all sources.  Many hearing the author's demonstration of 3D drop their jaw, go wide-eyed, and express a desire to have its realism.

While for listeners who have heard them compared, 5.1/6.1 is spacious and enveloping; however hybrid žPerAmbio 6.1.10Ó - at a žcostÓ of one decoder and 4 or 5 speakers - transports the imagination.  For some, given the choice, these improvements will justify their added cost; for others, they might not.  Most say that the listening area is large -  almost anywhere within the speaker sphere except near a speaker - and is enjoyable-sounding even outside the Ambisonic array!  At the Ambiophonic focus, the accurate front stage image compels a žbeliefÓ that you are there.

Fig.11 - Multi-format control room monitoring (showing 7 of 18 speakers +2 subwoofers) is switchable between PerAmbio 6.1.10 and ITU 5.1/6.1 (no need to change listening position as in Fig.1).  The common žSweet spotÓ accommodates two within the control roomŪs reflection free zone (RFZ).  In addition to Periphonic 3D surround demonstrations, this facility produces compatible 5.1 and PanAmbio 4.1 mixes, including evaluation CDs available to audio engineers at www.filmaker.com.

Listening Results - Transformed ITU 5.1/6.1

During PerAmbio 6.1.10+ recording or post-production, the recording engineer has chosen one of six encoding modes in order to optimize 2D reproduction in ITU 5.1/6.1.  This choice has no affect on ultimate 3D full sphere replay, where decoding mirrors the chosen mode.  No decoding is required for 2D 5.1/6.1 replay and, from informal listening tests, the results are uncompromised in 2D in terms of perceived quality in comparison to many commercial recordings made or remixed for 2D-only.

In demonstrations at the authorŪs studios using experimental recordings, made both in concert halls and studio, results in standard ITU 6.1 (discrete Surround-Center speaker) are gratifying.  Played on common studio and consumer speakers (see Fig. 11), several reported observations follow, first when played in ITU 6.1, then when the recordings are reconstituted in 3D.  The test recordings are simply excerpts with no EQ, level compression, or reverberation added:

Excerpt #1: Pipe organ with Trumpet rank directly overhead and Antiphonal division directly back

This excerpt is a staple of the authorŪs PanAmbio 4.1 demonstrations, where the 32 foot pedal (down to 16Hz) is redirected to the ž.1Ó LFE channel.  While giving a significant improvement over stereo in a more spacious, enveloping, and integrated room impression front-to-back, 6.1 also reveals that the antiphonal division is behind you.  The Trumpet that was overhead is merely all-around (cf. 3D below).

Excerpt #2: Harp concerto with small orchestra (ten players) in 500 seat hall - microphone array at greater than critical distance, raised and tilted -30°.

As the author has come to expect, initially this recording was observed by another recording engineer who was present during the session as žover-ambient, seeming to cause some loss of front focusÓ unless the three surround speakers were reduced -8dB and the room impression was žcorrect.Ó  After a weekŪs time, he found the ambience žborderline too muchÓ unless reduced just -3dB.  He rationalized it was his long-time conditioning to 2-speaker stereo.  (See different reaction in 3D below.)

Excerpt #3: Guitar concerto with small orchestra (ten players) in 500 seat hall - array at less than critical distance, at player level and not tilted (horizontal).

This example is usually played in informal listening tests beginning in 2-channel stereo, then switching mid-excerpt to 5.1/6.1.  The reaction is nearly universal - even žuntrainedÓ listeners notice an obvious difference, describing it as žsound wrapping aroundÓ them or words similar.  (A stronger reaction is universally reported in 3D, below.)

Excerpt #4: Oboe, piano, & trumpet in 500m3 studio with array at critical distance and level with players.

This experimental recording, played first in 2-speaker stereo, was received by the musicians who performed it as žacceptableÓ as a demo for promoting their ensemble.  However, in 6.1 it garnered more interest, the reported effect being of replacing the listening room with the studio where they recorded.  They became interested in what was required for a home theater system to play 2D surround.  (See more dramatic reaction below in PerAmbio 3D.)

PerAmbio 6.1.10+ 3D Informal Listening Results

The same recordings were then played in PerAmbio 6.1.10+ using the minimum layout of 10 speakers, plus two subwoofers.  Reconstituted from dual-format recordings, the listening results in 3D are identical with the original PerAmbio 6.1.10+ replay - i.e. the system is lossless - so priceless musical performances recorded using PerAmbio 3D are žfuture-proofÓ when released in this 2D/3D system.

PerAmbio 6.1.10+ uses 10 or more speakers to envelop the listener in natural sound.  The author and his colleague Angelo Farina have demonstrated this hybrid layout in Parma, Italy and Bethlehem, PA USA (Fig. 11) using modest speakers for ambience.  For nearly all 3D listeners, hearing sound from overhead and underfoot from a recording is astonishing, as though the ceiling, floor, as well as walls have been yanked away!  Musicians express that PerAmbio is just what they heard during in the concert hall or studio and so is just like žnormal hearing.Ó  The leap from 2D 5.1/6.1 to PerAmbio 3D reproduction of otherwise identical recordings is like going from stereo to 5.1.

Again, the 3D recordings have no EQ, level compression, or artificial ambience added.  Engineers involved in the sessions and listening tests say the natural impression makes these conventions less needed.  They also remark that PerAmbio seems to obviate most need for spot and room microphones, along with their complexity and possibly conflicting imaging.  Here are some of the observations (compare with the 2D comments above):

Excerpt #1: Pipe organ with Trumpet rank directly overhead and Antiphonal division directly back

While giving a žvagueÓ sense in 2D (above) that the antiphonal division was behind, in 3D it is clearly in back.  Surprisingly, the ceiling-mounted Trumpet is now directly overhead!  The organist - a Julliard-trained and experienced recording artist - upon hearing the recording went visibly slack-jawed in wide-eyed amazement.  (He has expressed interest in recording an album in PerAmbio 3D.)

Excerpt #2: Harp concerto with small orchestra (ten players) in 500 seat hall - microphone array at greater than critical distance, raised and tilted -30°.

Paradoxically, this recording, observed by another recording engineer as overly ambient in 2D, seemed less so once height cues were added in 3D.  He rationalized that the more žnaturalÓ the reproduction (3D), the more acceptable surround is.  One conductor who participated in listening tests expressed the strong desire to have a system for his own home theater - and pressed the author for an estimate of žwhen?Ó

Excerpt #3: Guitar concerto with small orchestra (ten players) in 500 seat hall - array at less than critical distance, at player level and not tilted (horizontal).

This example is demonstrated in informal listening tests beginning in 2-channel stereo, then switching mid-excerpt to PerAmbio 3D.  The universal reaction is elevated with respect to 2D above - both žtrainedÓ and žuntrainedÓ listeners [19] expressing variously that they felt if they closed their eyes, the listening room had been žerasedÓ and replaced by the concert hall.  One audio engineer who was present during the recording called it ža sound holograph.Ó

Excerpt #4: Oboe, piano, & trumpet in 500m3 studio with array at critical distance and level with players.

While hearing this recording in 2-0 stereo the musicians thought it just average and in 5.1/6.1 a bit more enthusiastically as sounding žaround youÓ (see above), now in PerAmbio 3D, they discussed with great enthusiasm and at length how natural and uncolored their instruments sounded.  The oboist immediately exclaimed žFinally, itŪs my sound after all these years.Ó  (The author takes this to mean žThis replicates the sound I was making while performing.Ó)  For their leader, minor žmistakes [in performance] now seemed trivial,Ó transcended by having captured the underlying feeling of their performance that was now žmore intimately conveyed,Ó having žengulfedÓ them.

ITU 5.1 Recordings Compatible with PerAmbio

The PerAmbio 6.1.10+ layout enables users to enjoy legacy 5.1 recordings and movies.  By simply moving back in the listening space by 26% of the speaker diameter, the user is able to hear ITU 5.1 with speakers at the proper angles specified in the standard, as illustrated in Fig. 12.  Side speakers may be used for a form of 7.1 listening, as shown in Fig. 13.


Fig.12.  For 5.1, listeners sit back 26% of the speaker diameter, where angles are at ITU standards (requires DSP changes in levels/delays).


Fig.13. The four side speakers provide for a form of 7.1 listening.

<< Previous Page | Next Page >>

Article Pages 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8
 

^ Back to Top ^