|The Science of Domestic Concert Hall Design|
by Ralph Glasgal
Guide to Ambiophonic Home Theatre
This guide was written to make the Ambiophonic approach accessible to those who already own a multichannel (home theatre) system based on a surround sound processor or receiver.
The core of Ambio is essentially the highly beneficial cross-talk cancelation (XTC), preventing a loudspeaker's output from reaching the contralateral ear. XTC can be implemented in any audio system either electronically, mechanically or a combination of both. In addition, XTC is highly compatible with direct playback of stereo and multichannel sources, as well as of up-mixed playback via Dolby Pro Logic II (Lf,Rf, Ls, Rs; 4.1), IIx (6.2, incl surround back, Lsb, Rsb) and IIz (8.2, incl front height, Lfh, Rfh). Note that no center channel is used with ambio, as the 2 Ambio front channels (Lf, Rf) are placed at merely ~20° from each other.
Elsewhere on the Ambiophonics site you will find a purist example of multichannel Ambiophonics by Ralph Glasgal: 2 front and 2 rear channels with XTC (i.e. PanAmbio), with in addition synthetic "surround" sounds obtained by convolution of the front stereo channels with separate impulse responses measured from a desired acoustical space. This indeed yields a very believable "you are there" experience. However, it is not trivial to setup, as it requires either unobtuanium convolvers or complex computer-based processing. Further, it is limited in that it dismisses recorded multichannel information beyond 4 channels (and sometimes even beyond 2), and is limited to the match between the auditory scene one convolves from (e.g. impulse response of the jazz club your multichannel recording was made may not be available). Last, outdoor events may lack such acoustical "enclosure" altogether and so then there's no means to obtain surround sound without a direct surround recording from it.
Hence, it is desirable to also be able to incorporate Ambio with the much more popular and off-the-shelf surround processors and receivers and with all the information contained in multichannel recordings (whatever their short-comings may be). The author has experimented for several years with his Ambio home theater and is convinced it provides an order of magnitude greater realism than standard surround sound, as well as than 4-channel (Pan)Ambio, which in itself is a highly noticeable step up from regular stereo.
Introducing XTC in your surround system
The assumption remains that the target setup will not be a computer-based system, but one based on an advanced surround receiver or processor like one from the series by Onkyo (PR-SC885 through PR-SC8809), by Integra (DHC40.2 through DHC-80.2), Marantz (AV7005 through 8801), etc., but even old 5.1 receivers will do quite nicely if not as well.
There are two easy ways to introduce XTC, acoustically or digitally. Either way, remove your center speaker and place your front pair at about 20° from each other. If using acoustical XTC, you'll need to place a barrier in front of you. This should be within a foot of your head, and extend out towards the speakers by at least ~5 feet. Ideally the barrier is absorbent for superior XTC of high frequencies (I made many measurements, being inspired by Don Keele's work). As discussed elsewhere on the site, try it temporarily with a mattress if you have no professional absorbent materials like Owens Corning 703 or 705, or an absorbent room divider (which I use). Start listening with just stereo output using a stereo symphonic or jazz piece and notice that the instruments extend well beyond the speakers and even beyond ±60°, and that a realistic depth is obtained. At the same time, it will sound somewhat "weird", which is simply because you have grown accustomed to standard stereo sound with its comb-filtering etc. to which we too adjust. So - dishabituate your brain from stereo. Within a few hours or less Ambio becomes the new normal and stereo will not only sound "weird" but also spatially more limited and generally dull.
If an acoustical barrier is not possible (though it still yields best sound imho), then digital XTC is needed. The current standard is the RACE-based DSP (see ambiophonics.org for details), which includes recursive XTC and sounds quite good after proper setup, close to the sound using the acoustical barrier, but not as good at XTC at high frequencies (where ILD rules). Simply insert e.g. a MiniDSP miniAmbio (~$125, plus a 5V DC power supply) between the processor's L&R front output and the power amp. After adjusting for distance and angle you will hear essentially the same effect as with the acoustical approach. Nice thing is that with DSP a monitor remains unobstructed from view (but see a later section). Ideally, surround processor manufacturers would incorporate RACE XTC DSP in their future offerings.
Now is a good time to re-equalize your system, as either approach will affect the spectral balance of sound arriving at each ipsilateral ear. I highly recommend Audyssey MultEQ XT32 and Sub EQ HT, which yields psychoacoustically flat responses and is a significant improvement over older versions by the same company. In general, re-eq after changing anything about the XTC configuration. Listen to a stereo source again, and engage your preferred surround mode. I think you will like what you hear.
PanAmbio or surround backs?
Assuming you have given yourself some time to get used to your now cross-talk free frontal playback and decided you prefer it over standard stereo, you should also consider implementing XTC for the rear surrounds, located close to midline (±~160°), using the same techniques as for the front channels. For ambience only surrounds XTC is not needed, assuming they are placed at ±~100°. You could stop there and enjoy great realism.
You can also consider the PanAmbio approach, routing either the front (Lf, Rf) or surround (Ls, Rs) channels to the surround back speakers, making the setup symmetrical front-back. I personally do not prefer this approach over Dolby PLIIx or PLIIz derived rear-surround envelopment in my current setup for most multichannel (but not 2-channel) sources, but it's close and integrates very well with the other channels. Experiment and decide for yourself. Its easy to do with XLR female to two XLR male splitters (or RCA male to two RCA female).
Wide or High: reverb
The more channels the greater the sense of envelopment, although this does asymptote. Modern processors provide height and/or width channels. The L&F height channels (H) add realism in making the sound stage more spacious and lifting the proscenium a bit. The L&F width channels (W) can "tickle the pinnae" by providing side cues at an optimal ±~55°. Currently, the width channel is accessible via DTS Neo X and Audyssey's DSX coding, but requires the center channel (C) in case of DSX, which we just got rid off for XTC (Audyssey should dump this requirement a.s.a.p.). The center channel could however be enabled and mixed into the Lf and Rf (F) at -6dB using an external mixer, which works fine¹. For 9.2 channel processors engaging W requires giving up either height or surround back, not so for 11.2 processors. Alternatively, we can forgo Audyssey DSX and derive wide channels on our own by a reverb unit. The Lexicon MX400 provides realistic surround reverb that is parametrically adjustable. Many configurations are possible, outlined below.
XTC is assumed for the F and SB channels. Of course one can add reverb or convolution-based channels to these configurations as desired, including as height channels. Especially a rear-height source may add envelopment to the more complex setups. The author often uses the PanAmbio configuration with reverb wide surrounds for 2-channel material over Dolby PLIIx/z. As said, switching between these configurations can be very easy (though eq may be needed).
Video equipment when using an XTC barrier
When using an acoustical XTC barrier a monitor can be placed above or to the side of the barrier. Tall metal storage racks (available at DIY stores) can help, as well as small side tables. This is fine for menu control of e.g. a media server or surround processor's menu, but not great for media content like concert videos or movies.
The best solution turns out to be to use head-mounted displays (HMDs). The HMD rests on your face like a "visor" and has a high-resolution screen placed about an inch in front of each eye. This results in a huge subjective "screen" size, allows for 3D viewing, and gives a centered perspective for each audience member. It takes a little work to place the HMD correctly so that the entire field of view is in focus, but once that's figured out it is easily repeated. Two models are available that the author has had great experience with.
First the Sony HMZ-t1 (~$700) and t2, which use two 1280x720 OLED displays. The t1 requires the tricky removal of the integrated headphones and associated ridge (to preserve your HRTF), whereas the t2 (~$900) is good to go as-is.
Second, the Silicon Micro Display ST1080 (~$800) uses two LCoS 1920x1080 displays and requires no modification. The higher resolution of the ST1080 is quite noticeable and desirable. However, screen brightness is not uniform and can be distracting initially. The Sony HMDs come with an interface box that will pass-through the HDMI signal to a monitor for easy HMD-less viewing (e.g. audio only sessions).
The author's reference setup
Files: mainly 24/96 2-8 channel flac
XTC: 4" absorbent barrier in front, above and behind the listener
This work would not have been possible without the generous, insightful and energetic discussions with Ralph Glasgal, Alvin Foster, and Leslie Williams.