I am curious (on a basic level) how DSX creates extra channels from existing channels.
Maybe I shouldn't because I'm probably wrong, but I'd really like to take a stab at it: information that is the same in both a front channel and it's corresponding surround channel gets added to the matrixed wide channel, while information that is different is ignored. And somehow the difference in magnitude of that information between front and surround channels determines the magnitude of the information added to the wide. I wouldn't have a clue for heights unless it's a 7.1 soundtrack and uses front and surround back information.
Okay, now how does it really work?
I would also like to know how DSX is supposed to reproduce say a small room, open area, or large room. Is this achieved through the timing of the reflection speakers or am I completely wrong. Thank you very much. There seems to be quite the debate about how this can be achieved on one of the forums and I would like to be able to provide everyone with accurate information. Thank you
DSX has absolutely nothing to do with matrixing. It derives from our studies of how sound is reflected and the role of reflections on our perception of space. The two most important reflection directions are (1) the side walls and (2) the surfaces above the performance stage. Measurements in concert halls and other venues (small and large) have provided the basis for our research and for the algorithm that synthesizes these reflections. They must be delivered from certain directions and with specific frequency and time shaping to properly combine with the direct sound coming from the front three speakers.
So DSX really doesn't have any way of making a room sound smaller or larger dependent of the source material is this correct?
If this is true what would the argument for DSX with say someone that treats there walls in favor of no perceived reflection at all?
That's exactly what DSX does: it controls the reflections to make the room sound larger or smaller depending on the content. These are not the listening room reflections. They should be eliminated via room treatments and room correction software (MultEQ). These are synthesized reflections that come from the Wide and Height speakers in a controlled way.
Thank you very much for clearing that up for me Chris. Always appreciate the help.
I had another question I was hoping you can help me out with. I'd like to hold out on purchasing a processor so I can buy one capable of 11 channel playback at once but it sounds like that may be a while. I know denon is capable but I was more interested in using a processor with Sabre DAC's or at least from a company like Marantz. If I were to buy a Denon would it be able to take a 7.1 analog signal from say an oppo 95 bluray player and convert it to 11 channels or does it have to be a digital signal? Thank You
AVR manufacturers don't put ADCs on the multichannel analog input. That means that no digital process can be applied to those inputs. That includes even basic things like bass management and, of course, all Audyssey algorithms.
Hey Logan. Just my 2 cents, but in my opinion don't worry about the DAC's. I don't think you would be able to tell the difference in a blind test.
Regarding DSX Chris, are you saying that the technology does not really add surround sound channels as we think of surround sound, meaning the channels provide an anchor point for objects in space being at a certain position, but rather with DSX, the channels provide only synthetic reflection information to make the viewer/listener perceive they are in a larger room such as a movie theater?
I am not sure what your definition of "surround channels" is exactly. If you mean "ambient" then the answer is "No. DSX does not just add ambient information". Actually, it adds direct and ambient sound that is extracted from the content based on our study of the reflection patterns in all types of rooms. It is these reflection patterns that affect our perception of source width and depth.
There's some confusion about how DSX works and what it does and doesn't do due to the varying interpretations of some comments attributed to you. The comments are posted below:
Our sense of auditory source width is greatly enhanced by reflections from the sides. For these to be most effective they need to arrive from a specific angle and with a specific delay relative to the direct sound from the front. They also need to have a particular frequency response shaping in order to match our hearing preferences. Experiments in the literature have shown these effects quite clearly and we have also performed our own at my university lab for several years.
The information must be delivered from a discrete direction that is separate from the fronts or surrounds to be effective. So, the Wides play a key role in soundstage rendering.
If we ever got to the point where content was being authored for more channels then the Wides would also play a key role in filling the front-surround gap. We made an experimental recording of a Shakespeare play happening "in the round" around the audience. When the actor's voice appears in the right surround (there is no picture--this is all presented in the dark), and then he starts slowly walking towards the voices he hears in the front left, the result is chilling. He is in the room with you and you can swear that somebody just walked by you. When we then switch to 5.1 the illusion is gone completely."
Some folks have taken the last paragraph to mean that DSX produces some channel steering to the Wides providing better panning.
I interpreted that last paragraph to mean that you made a special non-conventional recording with 7 discrete channels (the 5 regular channels plus the Wides) to compare the panning vs the conventional 5 channels, and your conclusion was that when audio mixing conventions advance to the point of adding discrete width channels to the mix, the Wide speakers will be of even greater benefit with respect to panning.
At this time though, DSX simply reproduces simulated ideal reflections to create a wider soundstage. Is that correct?
Or is the other person correct in interpreting your comments to mean that DSX as it stands now, actually matrixes info from the various channels to create smoother front to rear pans?
Thanks for any clarification you can add.
There is no channel steering in DSX. We found that steering can cause artifacts in imaging and so do not use it. The improved panning in the Shakespeare piece comes from the fact that the biggest perceptual gap in a surround system is being filled.
The front speakers are too far from the surrounds and that leaves the entire side area open. It's the most difficult area to fill because of the imaging mechanism of our ears that relies on time and level differences of arrival at the two ears. The Wides fill that gap nicely and improve front-to-back panning for content that has it.
The panning improvement is greatest with discrete content (that currently only exists in our lab). However, there is improvement in 5.1 content as well. For example, in Ch. 2 of WallE, the robot pans from the left surround to the front. It is much smoother with DSX on because the Left Wide fills the gap.
Are you talking about surrounds at 90°?
Didn't you say DSX doesn't use any steering? How can panning be improved if the content that gets panned isn't steered into the wides (or is discrete content)? Or is that panning is perceived as being smoother (as a result of added spaciousness) while phantom source localization accuracy isn't really any better than before?
Surrounds should be at 90° if they are dipoles or 110° if they are direct radiators.
Panned sounds have to "pass" through the Wides and therefore the reliance on all phantom rendering is reduced. Of course, discrete content is better than non-discrete as will always be the case.
What's your reasoning behind the statement "Surrounds should be at 90° if they are dipoles or 110° if they are direct radiators."?
At what angle were the surrounds placed in the "Shakespeare piece"? What type of surrounds were used?
Dipoles are designed to be at 90° to the listener. That's because their null should point at the listener. Direct radiators should be placed further back to increase envelopment. This comes from the ITU recommendation based on extensive listening evaluations for the ATSC standard.
The Shakespeare piece used direct radiators at 110° for surrounds.
BTW, if you're interested there is a picture of my university lab in this article that appeared in the NY Times. If you click on the picture to enlarge it you can see most of the speakers. The photograph was taken standing next to the single Back Surround speaker. You can see the Left Dipole Surround up high on a stand, but unfortunately the Direct Surrounds are not visible (they are further back).
Thanks. That room is 18000 cu. ft.? Pretty large when compared to the average home theater that is around 3000 cu. ft. Do results from such a rather large space translate well to smaller home theaters?
Yes they do because the acoustical characteristics including RT60 and direct-to-reflected ratio have been engineered to match a typical living room.
Hi Chris. Does every DSX reciever have the ability to read the discrete material that you created in your lab? Is there any chance that either audyssey or another outside company will create DSX audio codecs to take advantage of DSX 11 discrete channels rather than deriving the sound from 5.1 or 7.1 sources?
There is no way currently to deliver discrete 10+ channel material to consumers. We are not in the codec business, but those that are will probably be working releasing formats beyond 5.1 in the near future.
"Yes they do because the acoustical characteristics including RT60 and direct-to-reflected ratio have been engineered to match a typical living room."
Aren't these concepts (RT60 and direct-to-reflected ratio) derived from (and are applicable only to) diffuse sound fields? A typical living room doesn't have a diffuse sound field so I wonder if the actual reflection pattern (delay, angle, level, spectrum) within an acoustically small room is a more meaningful descriptor of the sound field?
A typical living room is dominated by the reverberant field energy. This is the opposite of what happens in dubbing stages and movie theaters where the listener hears mostly direct sound and very little reverberant sound. So a translation is needed to bring the living room to the same conditions as the studio.
Not sure how your comment relates to my question how the USC room (19000 cu. ft.) translates to normal living rooms (~3000 cu. ft.)? The USC room seems to be treated. One would expect the soundfield in that room to be closer to a dubbing stage or a movie theater than a living room - larger ITDG, more diffusive.
No, the USC lab is not acoustically close to a dubbing stage. It has RT60 and reflection patterns much closer to a living room. We do treat some side wall reflections, but most of the treatments are for low frequency modes (tuned absorbers) and diffusion (ceiling, front and back wall).
Would be very interesting to see how these 3 types of rooms (dubbing stage/movie theater, USC room, 3000 cu. ft. living room) compare to each other when looking at a) common parameters like RT and b) more specific parameters that show reflection paths, delay, spectrum, level. To my knowledge there haven't been lots of thorough investigations yet.