John was kind enough to send me his 2-DVD set of classic rock demos, giving me the chance to listen carefully on my best system over a period of time.
First, a word about derived multichannel.
There have been many systems developed over the years to synthesize a soundstage covering all four, six, or eight channels from a stereo source. The only such methods that have earned respect from the "quaddies" (that's us) are those that used the phase relationships to expand the normal stereo panorama between left and right into an arc that extends from left surround, through the front channels, to the right surround. The first of these was the Tate "stereo enhance" setting, which used analog technology. Later, this same principle was employed in LCRS systems such as Dolby Prologic, and exists today in Prologic II Music (when set to create the described panorama).
I have always been a big fan of these Tate-based systems, at times almost to the extent that I've honestly wondered if all the effort and disappointment pursuing "true" multichannel was even worthwhile.
Good, derived multichannel has undeniable advantages:
1) It's cheap.
2) Unlimited supply of source material (anything two-channel)
3) Nearly always produces a decent discrete soundfield, the type preferred by MC enthusiasts
4) Does not change the original mix
These last two are crucial.
The hard fact is that, in my opinion, over half of all multichannel remixes ever done are bad, for one of three reasons:
1) Poor sound quality (worse than original stereo)
2) Poor channel placement (in rare cases, "overly discrete" or inappropriate isolation; more commonly, weak separation or isolation of instruments and voice
3) Bad mix (unsupportable changes from original, missing elements, inappropriate level changes, poor use of or changes to effects)
There have simply always been far, far too many cases where a multichannel mix does not measure up to the original stereo for one, two, or all three reasons. I honestly don't understand why this doesn't seem to bother many in the multichannel community much more than it does. It bothers me a great deal.
So, now that you know my feelings on deriving multichannel from stereo, how does PenteoSurround stack up?
First let me say that, as you might expect, the degree to which the process is successful varies widely from track to track, as it does for Tate and DPLII. Tracks like Boston's "Don't Look Back" make for a pleasant, diffuse field, but nothing you would pick out as a demo. "All Along the Watchtower", however, will turn heads as sounds come at you from every direction.
Then there's the Beatles track, "Hey Jude". Paul is center, piano dead left back, other stuff dead back right. Pretty discrete, right? These late 60's tracks seems to decode very nicely, what with their mostly mono source tracks. But what's in the front channels? Not much. Many such tracks end up this way, basically "triphonic".
In fact, with all these systems, it's worth asking: what exactly is supposed to end up in the LF and RF channels? Logically, anything panned somewhere between Center and LB should appear LF -- exactly where, I'm not sure. Diffuse sounds, like reverb, should end up at least partially there. But actually pinpointing an exact instrument -- maybe a (mono) guitar -- in a front channel really doesn't happen in any of these systems. The lack of a truly independent front soundstage is the one thing derived systems simply can't deliver. (But true MC matrix systems, like SQ, can.)
So now, in addition to the previous points for comparison with the real thing, we have a fourth question: how does Penteosurround stack up against existing derived-MC systems?
A few thoughts:
1) The isolation and localization of Center, LB and RB are near-total: as discrete as can be. This is a good thing as far as Center goes -- having vocals dead center is a definite improvement. For the surrounds, one might argue the high level of discreteness is less natural than Tate or DPLII, or "triphonic".
2) Availability of recordings is limited to those actually processed (by hand) by Penteo
So what we have is a more-discrete alternative to Tate or DPLII, (which may be an improvement, depending on material), but which carries the ball-and-chain of being limited to material processed and released by Penteo or its deputies. Obviously this limitation is nothing compared to the hardships of dealing with original multitrack sources, but it is still significant.
One last test I would like to perform is an A/B of John's demo tracks with the same stereo versions played through DPLII. It won't be fair, of course, because they won't have benefitted from John's phase correcting.
No matter how things turn out for Penteo, I want to thank John for devoting so much time and work to bringing multichannel music into 2007. He's exploring alternatives to deal with the many obstacles this paradigm faces, while attempting to develop something satisfying and practical. For that, I salute and support him.
If I were king, before allowing any classic album to be remixed in multichannel, I would have John prepare a Penteo version from the stereo, then play it for the remix engineer over good speakers. "Hear that?" I'd say. "If you can't do better than that -- LOTS better -- then let's just call it a day."