What gonk and sdurani are saying is that DTS-ES *is* Neo:6. At least DTS-ES Matrix (DTS-ES discrete is, of course, six discrete channels + LFE).

Even though you're pushing the DTS-ES button, on a 5.1 DTS track what is actually happening is that the Neo:6 circuit is being engaged to extract the sixth channel.

Interestingly, DTS-ES Discrete also matrixes the sixth channel into the surround left and right, in addition to providing it on a discrete channel. The decoder removes the matrixed info from the surround channels the same way it would with DTS-ES Matrix, but can do so more accurately because it knows exactly what should come out.

------------------
Matthew J. Hill
matt@idsi.net

Not quite as "discrete" as most people believe it to be. Good point Matthew, and one that is often overlooked.

Best,
Sanjay

They do it that way so that 5.1 systems will get a "phantom" rear center, the same way DD-EX will have when played back on a 5.1 system. If it were true discrete 5.1, then the sixth channel would simply be lost if you didn't have the hardware to decode it.

Theoretically, if they do it right, it should be no different than "true" 6.1 discrete when played back on appropriate hardware, because removing the sixth channel from the two surrounds is simple math once you know what the channel actually is. Kind of like the way RGB video can be "exactly" determined from component video by simple math. There actually are six discrete channels in there.

------------------
Matthew J. Hill
matt@idsi.net

Matthew,

The "simple math" argument can be used for properly encoded ES Matrix and EX mixes too; after all, you know what the sixth channel is and therefore it is this known quantity that is being subtracted from the two surrounds. This 'known quantity' also means that the matrix derived ES channel will be pretty much as stable as the discrete ES channel.

As discrete as the sixth channel is in ES Discrete, it achieves this result at some expense to the side channels. Yes, the ES Discrete system is a fixed passive matrix rather than a dynamic active matrix, but it is a matrix nonetheless and as such prone to audible side-effects. There's no such thing as perfect matrix decoding. Still, since we're using a known quantity for the sixth channel, there probably won't be any of the typical artifacts we hear with dynamic active matrix decoders: i.e., pumping, leakage, etc. However, there are other possible problems. Follow:

After an ES soundtrack is mixed (with the sixth channel info matrixed into the surrounds), it is then compressed and coded using the DTS CODEC. Since this is a perceptual (lossy) CODEC, when the subtraction process removes the sixth channel info from the left & right discrete surround channels, the coding artifacts (that were masked by the presence of the sixth channel info) don't cancel out perfectly and can be audible just like matrix decode errors are audible. Also, because of the perceptually coded nature of ES Discrete, many of these audible errors may actually be more unpredictable than with a standard matrix system; i.e., you would never hear these kind of problems even with old Pro Logic decoding (no lossy perceptual coding). The concepts might be simple but, unfortunately, the math isn't.

Now, these are all problems that can occur with ES Matrix and EX. So discrete and matrixed sixth channels would seem to be on fairly equal footing, whether it comes to the stability of the channel itself or the extraction of that channel from the surrounds. But there's still one interesting difference when comparing ES matrix DVDs to ES Discrete DVDs.

One would figure that ES Discrete soundtracks with their additional discrete channel would have another, say at least, 16% more audio data compared to a DVD that has an ES Matrix only soundtrack. But that turns out not to be the case as the data rate for both is exactly the same. Which means that in order to have a sixth discrete channel, DTS has to reduce the overall fidelity of the audio. And this sacrifice in quality (not to mention the new hardware required for encoding and decoding) is done for what? So we can have a surround back channel whose marginal superiority (if any) may go completely unnoticed because said channel is placed behind our heads, where psychoacoustics show that our human hearing acuity is at its worst?

Best,
Sanjay

Well, DTS is already something like 6 times the data rate of DD, so I'm not sure if a 16% reduction in data rate per channel is going to cause any perceptable effects. Unless, that is, the DTS encoding algorithm is drastically less efficient than DD's.

I don't remember the exact data rates, and I know they can vary, but I think that DD is usually something like 384kbps whereas DTS is in the range of 2.0Mbps.

I hear what you're saying about the compression artifacts, and that's something that I'd not considered. But I'm not sure that it would be all that apparent. Both the rear channel and the surround channels are, in fact, audio, and summing two audio tracks results in an audio track. One would hope that DTS's encoding algorithm would be able to encode any reasonable audio track without intorucing perceptable artificats, especially at the data rates that it consumes.

The compression argument aside, though, you still have three true discrete rear channels whether they are represented as LS, CS, RS, or LS+CS, CS, CS+RS. A matrix encoder normally doesn't know what RS is, which is why it has to make psychoacoustic "guesses." This is what sets matrix algorithms apart from discrete audio. Once CS is a known quantity, however, then LS ans RS can be easily derived.

It is still directly behind your head, though.

------------------
Matthew J. Hill
matt@idsi.net

Matthew, Well, considering that DTS hasn't been demonstrated to sound better than DD despite having "something like 6 times the data rate of DD", I'd have to say Yes: DTS's codec is less efficient. I'm not all that sure how apparent it would be either. But my point was that the actual process isn't as simple as the concept may make it seem at first. Remember that the surround channels already have the sixth channel mixed in before going through the lossy compression process. This process uses perceptual coding to throw away data based on sounds it thinks we won't notice are missing, due to those sounds being covered up by other louder sounds as well as taking advantage of deficiencies in our all-too-human hearing.

The problem arises when, after the DTS stream is converted back to PCM, the sixth channel info is removed from the surround channels. What if it turned out that some of the sounds, that were tossed out during the compression process, were never noticed as missing because they were being masked by sounds in the sixth channel? And now those sounds from the sixth channel are gone and that missing data is "unmasked". Understand what I'm getting at? No question about that: you do have 3 discrete surround channels, something that the matrix versions can't claim. However, my point was to question whether any advantages of having a discrete sixth channel was worth the downside.

As you said, matrix decoders have "guess" unless they have a known quantity to deal with. In the case of ES-Discrete, the known quantity is a disctrete sixth channel. However, things aren't that much different for the ES-Matrix version: the sixth channel is encoded into the mix and decoded out of the same mix using a known scheme, without any guesses being used to extract that channel and then remove that info from the surrounds. It is this aspect specifically that allows for the ES-Matrix process to rival ES-Discrete when it comes to stability and sonics. No leakage, no pumping, none of the typical artifacts of using matrix decoders. So I don't see any big advantage to having a disctere sixth channel over a matrix one.

As for possible burdens: you need new hardware to take advantage of the sixth discrete channel. Even today, many receivers/pre-pros that do DTS ES-Matrix can't do ES-Discrete. Software-wise, as I said before, the data rate for ALL the other channels has to be reduced in order to accomodate the discrete sixth channel. To whatever extent that could be an audible problem, it is a problem that ES-Matrix-only soundtracks don't have to deal with. The Final Insult . As I said before, all that effort to get a discrete sixth channel and it ends up being placed where our hearing is the worst.

Look, I'd take a discrete channel over a matrix channel as much as the next guy, IF it is going to mean a real and audible difference. ES-Discrete may have advantages over ES-Matrix, but if those advantages are going to go unnoticed by listeners then what's the point?

Best,
Sanjay

Well, hey, I think we're both going around in circles now. And I think we're saying a lot of the same things: DTS-ES discrete would have been better if the rear channel had not been summed into the two surround channels.

Since it's not easy to make a direct comparison (know of any disks that are encoded in both matrix and discrete?) this is kind of theoretical.

I guess DTS really couldn't have done it any other way, though; in order for discrete tracks to play back on older hardware already out in the field, they had to make it backwards compatible. So, to avoid losing information in the stream, they had to sum the rear channel into the surrounds.

What they really should have done is anticipate this situation and gone 6.1 from the beginning. They could have had 5.1 systems do a downmix. But they didn't.

Oh well.

------------------
Matthew J. Hill
matt@idsi.net

Matthew,

Yeah, we are kinda going around in circles. So I'll make one last clarification.

I actually don't mind that the rear channel has been matrixed into the two surround back channel; from my experience, proper encoding and extraction can yield incredibly clean results. What I was getting at was that the a discrete rear channel was a pointless effort because it offered no audible benefit over the the matrixed version. DTS may not have had any other way to do DTS ES Discrete 6.1, but my point is that they shouldn't have bothered in the first place.

More smoke and mirrors from a company that has a history of doing that. Sorry to be so blunt.

Best,
Sanjay