DTS Neo.X - Page 8

Yikes, that's an hour long, and the video keeps stopping.

If you managed to get through it, could you summarize the technical highlights, and what products will have it?

Thanks

That's what I understood as well. Matrix encoding wide channels in their respective fronts and sides is obvious, since extraction is a simple 3-2-3 matrix and those sounds will phantom image at the correct location if you don't have side speakers.

Since height info is matrixed partially using both rear channels, I wonder if it will be encoded in stereo (centre + left rear, centre + right rear) or mono (centre + both rears). Maher said that they'll have an upcoming white paper, so I guess we'll wait and see. I suppose someone with a 7.1 encoder and an Onkyo 1009 could figure it out.

Exactly.


Lh is encoded into L/Lb; Rh in R/Rb. If L/R and Lb/Rb are stereo, the Lh/Rh will be, too.

I know you had hands-on experience with Neo:X height decoding when you had the 1009, but Maher said that height info is matrix encoded into C+Lb+Rb (as SoundChex also mentioned).

http://www.denon.eu/3D-edition/en/

AVC-A1HDA and the AVP-A1HDA will be updated.

I guess I'm referring to a different section of the chat. Did I miss it?

I listened to the MP3 starting at 38:30. Fred says, essentially: >>in the DAW I take the wides and pan them phantom between side left and front left. Height channels are panned exactly between front and rear-surround.<<

In the recent WSR article on Neo:X:

No mention of the center channel in either case. For good reason. Mixing two additional height channels into the center speaker would not be a good place to encode them: A) the separation will be lost, and ambience and spatiality need decorrelated signals to be convincing. And B) it would potentially cloud or compete with the dialog.

That's bizarre. I could have sworn I heard him mention the centre channel in relation to height encoding. Apparently so did SoundChex. Maybe inception really does work.

From the Fred Maher Podcast we "know" that DTS will 'encourage' movie makers to author '11.1 channel pseudo discrete' BD soundtracks as [backward compatible] 7.1 DTS-HDMA with front wide and front height content matrix encoded. We might expect that any such soundtrack will play back correctly in 11.1 and three 'flavors' of 9.1 when upmixed ['matrix-decoded' . . . ?] on any DTS Neo:X equipped AVR . . . but I'll be interested to hear how effectively the 'hidden channel' content is revealed by such 'competitor' upmix processors as might be found on an AVR that is NOT equipped with DTS Neo:X. Will 11.1 matrix encoded DTS-HDMA soundtracks 'play well' with DPLIIz, Audyssey DSX, and Yamaha CinemaDSP post processors, behave 'neutrally', or even interact adversely?

[Plus, I'm sure we're all wondering how well QLS 3D processing in the forthcoming 12.4 channel Lexicon MP-20 will upmix a DTS-HDMA "11.1-in-7.1" soundtrack by 'extracting' five height channels!? ]

Roger is correct; Fred said 'front' and NOT 'center' . . . which I found ambiguous in this context. With the L|R Mains already convoluted by matrices for the LW|RW channels, it seemed unlikely that they could be further complicated by the [3rd] matrix for LFH|RFH content, given that the extraction process would have to support four "two channel Neo:X upmix" configurations, to 5.x|7.x_Standard+2xFrontWide and 5.x|7.x_Standard+2xFrontHeight, in addition to two "four channel Neo:X upmix" configurations, to 5.x|7.x_Standard+2xFrontWide+2xFrontHeight. Hence, my 'center' paraphrasing . . . but with some uncertainty!?

Quote:
The algorithm determines the correlated signal present in any two channels and
creates with that signal one channel....between the front and rear channels the correlated signal is designated height-left and height-right.

That would seem to make misplaced height effects possible, like voices.

I know Fred mentioned "any two channels" but he did not mean every possible combination. Unlike DSX, Neo:X will not "false" the dialog into the height channels. The only way it will get there is if it is mixed into both a respective front (L or R) and back (Lb or Rb) pair.

Yes, for at least two flavors of 9.1 (7.1+heights or 7.1+wides). What's the third flavor?

I'm guessing the other processes will ignore the Neo:X encoding. They do not think the way Neo:X does (none of them look for correlation between front/side or front/rear). It should be benign, acting the same as any other 7.1 content.

I'm wondering how well QLS 3D will decode on multiple levels, including stereo!

One height + one wide?

5.x_Standard+2xFrontWide+2xFrontHeight (DTS refers to it as 9.1 Front-Height & Wide)

Afterthought: On hearing that the LR|RR channels are involved in the 'matrix encode process' for LFH|RFH content, it seems to me that all hypothetical|future "Neo:X encoded" soundtracks might have to be authored as DTS-HDMA 7.1 . . . leading to the possibility that "9.x stemmed" soundtracks mixed for 5.x_Standard+2xFrontWide+2xFrontHeight may actually 'upmixed status display' as 7.x_Standard+2xFrontWide+2xFrontHeight . . . but contain NULL LR|RR content after Neo:X expansion...?!

Which raises a fascinating question about whether the best way to address sound issues in a "non axially symmetrical" listening environment might be to allow 'some future version of Trinnov room correction' to add unpaired off-axis speakers.

It was a joke.

Nonetheless, my response was serious.

Actually I was mistakenly thinking of the heights actually having a significant elevation, like if they were to the sides, not front wall placement like DTS shows.

Another thought following the Fred Maher Podcast on Home Theater Geeks:

It seems to me that the matrix encode|decode mechanism Fred Maher described for the Front Wide channels allows for the full content of (say) the Left Main channel (found in the 5.1|7.1_Standard authored mix on a BD) to be 'relocated' into the (extracted|inferred) Left Wide channel (output in a 5.1|7.1_Standard+2xFrontWide Neo:X upmix). That should make it possible to create, either intentionally OR unintentionally, some great FULL POWER pans across all FIVE front speakers, e.g., LW => L => C => R=> RW.

I don't know what are the suggested 'performance characteristics' for speakers to be used as Front Wides in Audyssey DSX post-processing, but perhaps those to be used as Front Wides in DTS Neo:X post-processing should have a recommendation for similar power handling capabilities to the L|R Mains [after Bass redirection is considered]...?!

Yes, I can confirm that based on my tests. It completely steers equal signals in L/LS to Lw.



Yes.


Same as any other speaker. Think of it as just one more surround channel.

http://www.avsforum.com/avs-vb/showt...2#post20901062

Room done. Now I just want a new AVR/prepro that can give me Audyssey DSX and Neo:X. LFC would be nice too.

Currently using my old Denon AVR4810. Have to say that DSX does a stand up job panning sounds across the wide to side surrounds seamlessly. Hope DTS Neo:X is even more impressive.

Neo:X does wide panning correctly, and does not muck with the surrounds along the way. Whether that sounds more impressive is a matter of opinion.

Roger:
I wondered if, based on your own testing, you think the 'performance' of the Neo:X expansion process might be more sensitive to 'correct' speaker placement angles [i.e., L|R Mains 'near' ±30°, LW|RW 'near' ±60°, and LS|RS 'near' ±110°] than is (say) DSX...?

No. I would not think so at all. What did you have in mind?

It occurred to me that, consequent to our prior discussions, there exists the possibility for an 'encoded' DTS-HDMA "11.1-in-7.1" soundtrack 'to circle the front of the room', stepping from (e.g.) LS to LW to L to C to R to RW to RS., at (say) a constant angular velocity. The intervals between each speaker are "nominally" 50°, 30°, 30°, 30°, 30°, and 50°, respectively. Obviously, even on a 5.1 configuration, a speaker 'angular placement error' will have the effect of speeding up or slowing down the rotation at [some successive] intervals. Perhaps I'm just imagining the worst, but it seems that Neo:X matrix encode+decode process we've heard described allows for the possibility of replacing (say) a 'moderate' LS-to-L velocity vs. L-to-C-to-R velocity discrepancy with a 'more offensive' LW-to-L velocity vs. L-to-C-to-R velocity discrepancy through placements of the LW closer than -60° and the L Main wider than -30°...?!

If the soundtrack was mixed in a 5.1 setup, then let's assume that the mixing engineer regulated the panning such that the angular velocity from Ls to L is what he wanted, and from L to C to R was what he wanted.

If the Neo:X wide speakers were placed to bisect the L/Ls included angle, it would be positioned at 70° rather than 60°. That would yield a perfectly uniform angular pan through Ls-Lw-L. If the Lw speakers are at 60°, will the 10° offset hurt the pan? Can a human detect that shift in angular velocity whilst being engrossed in the movie? Will phantom image localization be any better? And considering the wide speakers are off-screen, there can be no visual disparity to alert the viewer.

If one is concerned about all this, then DSX wides are a non-starter, as it takes L-R pans and a) accelerates their rate of trajectory, and b) places the hard L (or R) sounds wider than the screen. In cases where a corresponding on-screen image exists, there will no longer be accurate alignment. And don't forget the level and bass emphasis DSX imparts along the way, further altering the nature of the original source.

Gotta say thanks to all contributors to this discussion. This is why so many come here... to learn from those in the know! I'm waiting awhile till I take the plunge. Just recently got an Onkyo 7.2 that does heights or wides or rear that I enjoy playing with and it makes me want 11.2 now!

Attachment 222625

I am in the middle of a house build for which I designed a semi-dedicated HT. Here's a surround sound floor plan with 12 speakers and 2 subs. The front is basic DSX Wide (towers) & Height (in-ceiling). But on the back, I added 2 in-ceilings @ 135° off axis (or @ 45° off rear axis) such as Yamaha uses (Rear Presence). I tested my RX-V2700 with Front Presence and I liked it. Having these in-celing rear "Heights", I feel 1 single Back Surround (tower) would be adequate.

I am simply speculating of course, but I might just wire everything for this, since I'd hate to mess up the drywall after a few years to set-up newest surround formats.

I'd still wire for 2 Back Surrounds. But without those rear in-ceilings, 1 Back Surround means only 10 channels of amplification, which are 2 five-channel amps (think Emotiva XPA-5 or future XPR-5), where 11 channels would mean 3 multichannel amps (with the exception of the Onkyo 9-ch amp + 1 stereo amp)

Please share your thoughts!

 

Ronse2010B_HTplan9_DSX12.pdf 468.9287109375k . file

Have you looked at Report ITU-R BS.2159-1 (10/2010) Multichannel sound technology in home and broadcasting applications...? I see that your proposed layout 'bears a strong resemblance' to the Typical small room installation schematic (page 14, FIGURE 8)--except that you have added a pair of Rear Height speakers 'consistent with the Yamaha Rear Presence channel requirements'.

You might also note that some possible future '8 channel downmix' configurations "tested" for A. Ando's paper Conversion of reproduced sound field based on the coincidence of sound pressure and direction of particle velocity (pages 4 and 5) include retention of one or two 'bottom layer' speakers, placed below the screen . . . plus various other 'unusual' choices.

I haven't seen that, but I will read it asap. I love these idea's. Thanks for the links!

I go the idea for the single back surround from Tomlinson Holman 10.2 system before he and Chris Kyriakakis developed DSX. It was Chris who confirmed me by e-mail that they use 2 back surrounds (11.1) because that's what the AVR's are used to do anyway. The second back surrounds brings very little extra vs only one, except for the largest rooms.

With the 2 rear "Heights" I added, you also have complete control over height IMO. Imagine a scene shot from beneath the runpath of an aeroplane during take off... But you'd need discrete channels to have full effect.