Originally Posted by Roger Dressler
I like how you think, here and in many other posts.
I have no doubt.
I’m curious how, or if, you distinguish DSX processing from others such as PLII, Neo:6, or Logic7. I see DSX as very different from the latter, and have described PLII (and PLIIx) exactly as you have above: >>No new sounds are created beyond what exist on the soundtrack.<< So what’s different about DSX vs PLII, since they both create new speaker signals? DSX does so by creating new signals, much the way Yamaha DSP does. Yes, these new signals are directly traceable to sounds in the source--it doesn't toss in the ring of a mobile phone at random. Now I’m not equating the effect between DSX and Yamaha DSP, just the mechanism they use to create their effects.
DSX creates early reflections to simulate side wall reflections. These are a form of short, diffused echoes, and those reflections do not exist in the original recordings, especially if we are talking about movies or pop music.
In contrast, PLII uses various sum/difference ratios of the L and R input signals to create the new outputs. There are no means to perform room modeling, wall reflections, or reverb.
That does not mean DSX is a gimmick, but it is psychoacoustic trickery, well done though it may be. PLII(x) is also psychoacoustic trickery, and so is 2-ch stereo. Nothing wrong with that in my book. DSX just happens to have a tool chest that includes creating sounds that were not in the source.
This is included here because I believe this illustrates the above point. The issue of the unfocused center is not a result of a lack of speakers, as the 5.1 sources exhibit no lack of focus. It is the DSX processing, the adding of simulated early reflections from panned dialog that causes it.
How've you been? (I, too, have much appreciated your posts--and thinking!--elsewhere on AVS.)
Wow, I'm not sure I properly can respond to all the ideas you are bringing up, but let me try. First of all, while I've experimented with Neo:6 and, briefly, have experienced Logic7, I do not use these processing options. Mostly because I do not, in the end, particularly care for them--at least often enough to bother messing around with them, after their novelty factor dissipated--especially given the need to move speakers somewhat to optimize their contribution. (Well, in the case of Logic7, also because I don't own a Harmon Itl.-made processor.) PLII/PLIIx is an option I have used, and enjoyed, on much program material; I often have found it useful (or at least audibly interesting), even for two-channel source material (although not always--there is NO panacea for all source material, of course). And note that I DO still find 2-channel stereo--as opposed to 11.X stereo--still quite compelling. At least in the (now seemingly so needlessly restrictive) "sweet spot."
[Digression for the neophytes: "stereophonic," or "stereo" does NOT mean "two-channel sound," although we tend to think of it that way, given our long experience with analog home Hi-Fi; when Bell Labs coined the term in the '30s, from the Greek stereo
--"sound"--it was working with and recommending a minimum of three (front) channels, and an ideal maximum of an infinite (surround sound) number, but there was no then-practical way seen to deliver more than two channels, using only two record-groove walls. So, we were stuck with a less-than-ideal two channel "stereo"...Quad's entrance thirty-some-odd years later, notwithstanding...and now, in the digital age, we have 5.X, 7.X, 9.X, 10.2, 11.X "stereo").]
) "And now back to your regularly-scheduled programming," lol:
Regarding whether Audyssey DSX ADDS sonic information? I don't believe this is the case. I believe that it RE-PLACES (not "replaces") extant audio cues, but does not "create early reflections," to more or less quote you (minus the plurailty of the verb). But, I may be wrong. Let's parse a quote from Audyssey's Chris Kyriakakis, regarding "desirable" sidewall reflections, in a reverberent space:
"The most important direction for these reflections is from the sides and that’s what the Audyssey DSX Wide channels are designed to do. The algorithm looks at the content
in real time and extracts from it the cues that we perceive from optimal side wall reflections
. This information combines with the direct sound from the front and gives us an enhanced sense of soundstage width [emphasis added].
Clearly, the implication is that the information is extracted
from existing information before being steered to the Wides. So, this is more than mere identification of summed common sounds existing in, say, the LF and LS, steered to the LW speaker, which presumably does happen to some degree, in order to increase the front-to-back "fill"--but the implication most importantly is that the critical "reflection" information comes from the mix, and is not created out of whole cloth. Now, changes in the time/phase information, we see logically, must occur (in order to present it properly at the "moved" point source of the Wide location, and this is done assuming a properly, precisely positioned Wide speaker, i.e.
+/- 60 degrees from center. And, I use the term "point source" loosely, to refer to the new speaker placement position in the 3D listening space, not in terms of loudspeaker design/engineering). So, there is, technically a 'change" in the sonic information--time- and phase-wise, but this is ONLY because it is necessary
when being presented at a new
point-source--and it can be accurately changed for presentation because the new positions are "known" by the algorithms to exist at precise point-sources, assuming one has setup the Wides precisely. By this "change" in time/phase and simultaneous "change" in transducer position, the net effect is of nothing added!
[As an aside, I might also remind you/everyone that by removing the existing early-reflection info. from the fronts' presentation, for proper re-location at the Wides' placements, this increases intelligibility and sonic nuance of the direct, frontal presentation by the LCR array: "un-muddying" and mimicking or, rather, complimenting, what other Audyssey technologies try to effect in terms of the loudspeaker/room interactions. (Ah, synergy!)]
Unfortunately, Chris does immediately preface the above quote with a semantically troubling sentence: "But, what if we could recreate
the desirable reflections? Then, we can really feel more immersed in the scene [emphasis added]." This might be interpreted (I think wrongly, given the context) such that the word "recreate" seems to mean "synthesize." I do not believe this is what Chris means, given many other articles (most much more technical and complex) that I've read. I'm sure he means "select" and "utilize," here--i.e.
extract the existing desirable reflections from the mix and send them to the chosen Wide position; the context is about how and why previously-existing Audyssey technologies serve to minimize undesirable reflections that, e.g. muddy intelligibility and spacial cues, yet here is this whole new DSX technology that tries to use
reflection information: positively
. The proper inference, I believe, is that the former exist because of (undesirable, innacurate-to-the-mix) speaker/room interactions; the latter (desirable reflections coming from the natural reverberation in the recording space or reasonable efforts to simulate such on the dubbing stage) exist in the mix, and are simply "moved" to their proper positon in the spatial (and time) array--e.g. to the Wides. And, it's critical to note: this reflection information does not occur at the Wides unless it exists in the mix!--i.e.
it is not constantly present, but is content-dependent; if it doesn't exist in the mix, it doesn't get sent to the Wides. Again, nothing new is "created." We're just getting a better psychoacoustic presentation of existing information, according to the way the pinnae (our outer ears, for the neophytes) gather sound for our brains' processing of spatial-localization cues. And, frankly, despite the hugely effective and impressive computing ability of our current DSP SOTA (state-of-the-art), I don't believe it is (yet?) possible at all to effectively simulate, presumably via the Wides, a plausible, scene-dependent, intermittent reverberation effect, on-the-fly. If it ain't in the mix, I don't believe it CAN be created, both intermittently and wholly effectively, at any real-world cost. The extent to which an attempt would often fail--a reasonable presumption--would be the extent to which the suspension of disbelief and the scene realism would collapse--along with consumer toleration and use.
Here's the link to the Audyssey info. I've quoted--which, as you'll see, is a pretty facile "introduction" kind of marketing post by Chris, and is therefore not particularly precise or technical, which leads to its possible mis-interpretation by neophytes (present specific
company, of course, excluded). I'll try and track down more technical info. that has informed my understanding of how DSX works, over these last few years, but may not be able to locate it. Still, I'm reasonably certain that DSX does not add information not extant in the mix--e.g. does NOT function as the various DSP "added enhancements" do, including the arguably excellent Yamaha processing, which I also have experienced to good effect--certainly it's the best of its breed, but it's definitely a different breed from DSX, AFAIK. (The Yamaha DSP processing does
"add" info., I believe, if only according to the naturally-occurring spatial reverberation of actually-digitally-sampled, complex, real 3D spaces, and using VERY sophisticated processing chips and algorithms; still, it is an additive technology, AFAIK, except for its front Height processing, which is derived from the mix, IIRCP...correct me if I'm wrong!) There are a couple of DSX-related and/or closely-related-tech white papers floating around, as I recall, and I remember a pretty good technical description that manages to be both forthcoming yet still protective of IP secrets, written by somebody really good at presenting complex technical info. to reasonably knowledgable hobbyists--perhaps Tom Norton (of Ultimate AV
and Home Theater Magazine
), I'm thinkin', or at least someone bright enough to try and follow in his estimable footsteps...Oops, sorry, that link:http://www.audyssey.com/blog/2010/05...-audyssey-dsx/
Again, it's pretty basic, and not very precise; obviously designed to try to explain the DSX function to the neophyte.
If you have or find info. that clarifies or challenges my undersanding, by all means post it! I realize and admit the weakness of the quoted material fully to support my understanding, and regret that I couldn't find immediately-to-hand the info. that I've relied on over the years in forming my present--but aways evolving!--understanding (such as it is).
Again, nice to "see" you,
P.S. A little more clarification for neophytes, in this already lengthy and information-dense post: Audyssey DSX does NOT "create" 11.X channels of information. It only "creates" four channels (actually, "create" is a bad word, given my whole point, above, as it creates no new info. at all, but I'll use it for simplicity). Again, DSX only creates Wides and/or Heights. It does this from 5.1 channel source material only; to the extent that 7 main channels of sound are available, either discretely or, say, via Dolby PLIIx processing, these remain unprocessed by Audyssey DSX. These extra two channels, the rear surrounds, are simply passed through. Also, 5.1 is a REQUIREMENT as an input source for Audyssey processing. Two-channel stereo cannot be processed by DSX. However, again, if the five-channel mix is "created" by some pre-processing mode such as Dolby Pro Logic IIx from a two-channel mix, then Audyssey will further process this information. However, this pre-processing, if present, is NOT a function of DSX, and the end results are NOT, if deleterious and resulting from this pre-processing, layable at Audyssey's door. This is all complex, I know. But I hope this cuts down on a lot of obvious confusion and mis-information, around here.
P.P.S. IMO: Those of you who "voted" for 7.1, or who voted for 5.1 because you can't use rear surrounds or think they're relatively useless: You owe it to yourself to audition and consider a "7.1 Audyssey DSX array"--with the normal 5.1 array augmented by Audyssey Wides, instead of (most-often mono) rear surrounds. Not only is our ear/brain processing geared for and most sensitive to frontal-locational 3D sound cueing, this is where most of the sonic info. in multi-channel mixes occurs (duh), and where the dynamic effects of multi-channel realism are best enhanced. So, for far less money than a fullblown 11.X array would set you back--perhaps for no increase in budget at all, except for getting a receiver or pre-pro that offers DSX processing--you can experience 90% of the benefits of DSX, and eliminate the (relatively, compared to the Wides' contribution to the realistic soundfield presentation) useless rear surrounds. So, this is all to say that it's not, necessarily a choice between 5.X/7.X OR
11.X, but maybe a choice between FLAVORS of 7.X. With Audyssey being my preferred choice, obviously. YMMV!
P.P.P.S. O.K. Last postscript! (But I may be lyin',eh?!
) Here's Audyssey's recommended evolution of added surround-sound channels:
Beginning with 5.X > 7.X with the addition of Audyssey Wides > 9.X with the addition of Audyssey Heights > 11.X with the addition of Rear Surrounds. This is based on the relative merits of these additional channels in enhancing the realistic portrayal of 3D acoustic spaces, based on actual, tested/modeled human psychoacoustic processing strengths/abilities via the ear/brain enviromental interface. (Gotcha: hows THAT for a mouthful of jargon! Spent some time on--and had some fun with--that one, I'm tellin' ya!) There are diminishing returns after the addition of the Wides, but the Wides, at least, ARE seen as quite critical. Obviously, our "current" 7.X Rears are considered relatively unimportant to adding to realism of the soundfield--especially as Audyssey recommends placement of the Side Surrounds at a placement of 15 degrees behind the sideplane of the ideal listening position, not directly to the sides, as with the Dolby 5.1 model; so the rear "fill" is adequate, given the way our ear/brain processing functions with sounds in this space (which is not very effectively--for fight-or-flight responses, we need only to process a general location cue from the rear, as we will turn our heads to focus ears and eyes for accurate 3D resolution, only as needed...while we're running away from the T-Rex...). This explains why so many of us are perfectly happy with 5.1, and often find 7.1, according to the Dolby model, relatively less important. It IS less important--far less. Wides, however, (to a much lesser extent, Heights), are almost equally as important as the Side Surrounds...Audyssey would argue that the post-Dolby 5.1 model is EXACTLY WRONG in its priorities: it moved to rear surrounds first--the least important for accurate spatial cueing--and now, with Dolby PLIIz, to front Heights (presumably to Wides, next?!) So, the net effect for poor Audyssey, unfortunately, is that the current 7.1 and 9.1 (with PLIIz, not Audyssey) installed-base of consumers certainly is gun-shy of hearing about the merits of even more speakers--since what they've added to date has been bass-ackwards and relatively ineffective, at great expense. But, as above, if the proper prioroty is given to the additional channels, the scourge of diminishing returns is banished, or at least properly scalable in a predictible and rational way.
My work is either done, here, or I'm sure as hades done with it!