That rings true to me. When I was running the old Yamaha DSP1 for the sound field speakers I often found that the Dolby Surround setting was the most natural and satisfactory. Despite Yamaha's great efforts to bring me the reverberant sound of genuine concert halls, the silly trick of putting the out of phase material (basically) into the surround speakers seemed the most natural.
Out of curiosity how many of you listen to music while working at a computer?? When you listen there how close are you, where are the speakers and do you have an issue with a phantom image??
I have my pair off to the sides of my flatscreen and in the same plane about
2 ft away. No issues at all I get a solid phantom image.
Rob
I have my pair off to the sides of my flatscreen and in the same plane about
2 ft away. No issues at all I get a solid phantom image.
Rob
I have the same experience if speakers are directly to the sides facing me, and not too far away. (Say no more than a metre from me on either side) The experience is much like headphone listening, and binaural recordings even work to some degree. Sound localization of normal recordings is more or less inside the head, as it would be for headphones.
However this is not what I experience with speakers at a normal 2-3 metre listening distance but separated by +/- 60 degrees. In that case I hear the same thing Markus describes. If I'm exactly equidistant there is some sense of a phantom centre but it is rather vague and not well focused at all, and moving to the side collapses it towards the nearer speaker rather easily. There is also little sense of stereo spread through the middle with most sounds near the sides.
So yes, you can still hear a vague "somewhere in the middle" sensation in this case but it's nothing like the solid pin-point convincing phantom centre you can get with +/- 30 degrees or less that would convince you that a dummy centre channel speaker was active when it was actually turned off.
I think it would be fair to say that speakers directly at the sides (+/- 90 degrees) is yet another "mode" of interpretation by our brains (headphone mode if you like) which is not just a wider version of +/- 60 degrees in terms of perception.
Last edited:
The question of how does a system create a center phantom image... it's how does the brain create the image? I know that in the lower midrange it's mostly about timing comparisons; laterally, and in the upper midrange it's mostly about amplitude comparisons. So if both ears get equal amplitude AND no differential time delay, the brain is going to create a phantom center image, is it not? I'm still curious about the 800HZ - 2kHZ region that is apparently associated with front or behind us cuing. That region may be more important than I've been thinking (?)...
I assume it is tied to training through our lifetime of comparisons between what is presented to the two ears and what we see. "Same signal, same delay or phase: must be dead center in front of me."
I'm doing a lot of work with 3D projection these days and it is amazing how strong the effect is when done well. We are well trained that a certain amount of eye convergence equates to a certain object distance. This is coroborated by our perception of the object size vs. what we know its real size is. When we perceive a consistent world ("eye convergence parallel, distant focus, large object looks small, must be far away") then the effect is very convincing.
David S.
Our ability to localize vs. the intensity "pattern" of the driver + boundary.
As you physically approach the operating driver - the signal changes (predominately increasing in spl, but also changing in other aspects time/phase and the interrelationship with amplitude). This difference is what provides our hearing the cue necessary for direct sound to dominate (..and we improve on it with difference checking from our binaural system).
Note that this is not the "precedence effect", rather the precedence effect is an observation derived from this.
With an acoustic "model", say a person speaking at a door threshold with the door wide open and the listener looking at that person, the direction and nature of that direction is obvious (or non-ambiguous). Of course it includes not only the direct sound, but also variety of reflections (both near and "far" - in amplitude and time).
Now *alter* the model so that door obscures the speaker from the listener (..i.e. the door is in front of the speaker and the listener can no longer see the speaker). At that point there is a "break" in the sound gradient emanating from the speaker's mouth/vocal chords (on its path to the listener). Of course most listeners will still be able to ascertain that person speaking is still at the door threshold (i.e. still remaining obvious as to direction), but the nature of that direction is no longer as clear - or does in fact "drift" into ambiguity. (..our ability to convolve the sound isn't perfect.) If we just alter the model by leaving the door open, but having the person speaking facing away from the listener - then a similar (though less obvious) reduction in clarity occurs. That simple rotation, though diffracting around the speaker's head, breaks the sound gradient.
Note that a break in the sound gradient allows for reflections to really start "competing" for directionality and increase ambiguity, or even change directionality (largely depending on intensity) - assuming of course that their own gradient is broken. (..this is despite the fact that the multiplicity of reflections and their own pressure gradient is substantially different than direct sound.)
Going back to loudspeakers:
For those who listen to horns, particularly large format horns/drivers, there is often the stated observation that once you horizontally rotate the loudspeaker to the point where you can no longer *see* the opening of the driver (i.e. the driver's exit), then the image becomes diffuse and directionality takes a substantial "drop" in quality (..but the sound tends to be less associated with the loudspeaker). THIS IS JUST IN MONO.
STEREO reproduction compounds this.
If it's a "toe-in" method (i.e. 0 degree crossing in front of the listener), then center image, despite the apparent diffusion, is still fairly directional - this is because difference checking helps to alleviate the problem where the signal from the left speaker reaches the right ear with *less* of a "gradient fracture" (than the left ear experiences), and often the pressure level is higher at the right ear (than normal when compared to the left ear) - creating a mono emphasis.
If it's a "toe-out" method (i.e. 0 degree crossing behind the listener), then center image becomes increasingly ambiguous once you can no longer see the driver's exit. There you have the gradient fracture AND a *decrease* from a mono emphasis. Stereo center turns to cr@p at this point, and it becomes very ambiguous - to the point where many describe a "hole" in the sound at center.
This happens to most traditional loudspeakers as well, but to a lesser extent - largely because the loudspeaker experiences less of the gradient fracture due to its design. The smaller the diameter driver for a given freq., the less it occurs. Driver shape is a factor here - convex drivers (i.e. "domes") aren't generally going to have this problem (unless substantially "inset" into their frame or baffle), at least not until the loudspeaker is rotated so far that the baffle itself obscures your vision from the driver. HOWEVER, you also have the problem of baffle edge diffraction "peaking" - creating very near time competing pressure levels and gradients that are difficult to differentiate. (..part of this difficulty lies in the fact that intensities of both the direct sound and the diffraction "peak" are both high and physically near to each other, and part of it is because the rotational change places diffraction more "in front" of the direct sound - which decreases the chance for difference checking.) Of course again you are also altering mono-emphasis, and this is a strong factor as well.
Finally I should of course note that this is largely dealing with changes in intensity with distance. As a result - as freq.s increase (including sibilants and harmonics) - that we are more effected (..up to a point). Said differently the sound gradient is more difficult to perceive as we go lower in freq..
All of this can of course be tested by each forum member.
Correct, envelopment ques should come all around us, not just from +-30 degrees. And these signals should not be correlated with the front channels. Correctly implemented LEDE room will give some of those (uncorrelated)signals as late 'reverberation' from 2 +-30 degree front channels.
So does a PL2 upmixer - but I think it doesn't decorrelate signals effectively, narrowing the depth perception a little. There is a better method developed by John Usher, check his PhD thesis. I don't know if its implemented somewhere.
Similar/same research as implemented in a Yamaha soundbar.
Also check out the beamforming presentation by Ivan Tashev (from Microsoft!):
http://www.aes.org/sections/pnw/ppt/ivantash/loudspeakerarraysforfocusinganddiffusingsound.pdf
Very interesting stuff and I'm wondering if a much simpler implementation (no array) of the 3 beam scenario tailored to ones specific room is doable? This is DIY after all
My latest ideas for my surround sound extractor project is as follows: There will be a half circle of five speakers. I'll put an L-XR on the front and side left and right. This will largely cancel the L+R signal while maintaining a sense of stereo effect in each pair. The center speaker will get L+R, and will be driven by a VCA that will be driven by a rectified and peak detected difference of L+R, before and after the L-XR that drives the front L&R speakers. As you turn up the amount of L-XR on the fronts, thereby cancelling the amount of L+R in them, the volume of the center speaker will increase, thereby becoming the dominant L+R source (in the upper midrange anyway). The L+R may get a discrete delay, either positive or negative, relative to all other outputs (who knows what I will learn from that). The four stereo outputs will go through a Lexicon MX400 quad reverb (only $300 but very good sounding). When only dialog is happening, the four stereo outputs will have little L+R in them, so hopefully the reverbs that I might have set up for the two pairs of stereo outputs (front and side L&R) won't be too obnoctously active. It's experimental, but you've got to start somewhere. I also plan to have a bandwidth limit on the control voltage to the VCA of 100HZ to 7kHZ, so bass and high treble won't affect the up to 3dB or so of "steering" of the center speaker. Compared to the nothing fancy version of this (no VCA), I hope to achieve at least another 3dB of separation between the center signal and the front or side L&R signals, which in addition to the non-fancy alleged 3dB of separation, may be just enough for the whole thing to be worth it (?).
Feel free to criticize this. Anyone been down this road?
I was originally going to steer the side outputs until I read a paper by David Griesinger where he pointed out the annoyance factor of total dB variance in the room, with incomplete steering. Complete steering, as in Dolby ProLogic II, monitors all 5+ outputs, maintaining a constant amount of dB SPL into the room, regardless of what steering is happening at any given moment. I decided to limit any amplitude modulation (steering) to just the center speaker in an effort to minimize this annoyance.
Feel free to criticize this. Anyone been down this road?
I was originally going to steer the side outputs until I read a paper by David Griesinger where he pointed out the annoyance factor of total dB variance in the room, with incomplete steering. Complete steering, as in Dolby ProLogic II, monitors all 5+ outputs, maintaining a constant amount of dB SPL into the room, regardless of what steering is happening at any given moment. I decided to limit any amplitude modulation (steering) to just the center speaker in an effort to minimize this annoyance.
Seems very interesting, thanks for sharing.
But the descriptions confused me, it's hard to visualized. Would you please explain a little more, with numbering or diagram?
And, how do you make those signal sums and subtractions? Any circuit, or simplied block diagram to share? Or it's a commercial product?
Thanks again.
But the descriptions confused me, it's hard to visualized. Would you please explain a little more, with numbering or diagram?
And, how do you make those signal sums and subtractions? Any circuit, or simplied block diagram to share? Or it's a commercial product?
Thanks again.
You need the array to get a tight beam. How mutch DI is enough and to what frequency? DI of ~12 is achievable with a horn/waveguide at high frequencies, or you could use a large(r) fullranger aimed at sidewalls.Very interesting stuff and I'm wondering if a much simpler implementation (no array) of the 3 beam scenario tailored to ones specific room is doable? This is DIY after all
Ofcource this non-ideal solution gives surround channel leakage from front, but it could be quite 'pleasant' sounding. Isn't this a little similar to a stereolith?
I was thinking the same thing (Stereolith). About the only practical diy approach would be to use cancelation to send broadband nulls in the listeners direction and get enough energy to bounce off the side walls for left and right.
I'm not sure what the point is in the end. The intent of the 1 Limited is that a customer can hang a single speaker and use electronic setup to steer 5 beams of sound around the room. It is really for someone who is balking at placing multiple speakers. The quality can only approach the 5 speaker quality, not surpass it.
A lot of technical sophistication and cost to avoid putting up the proper system.
David S.
Last edited:
Hooley writes "... to adequately steer very low frequencies requires quite large arrays ... which are impractical in most domestic environments. Here DDA [Digital Delay Array] tacitly uses the brain’s own psychoacoustic tricks, by delivering the lowest frequencies directly to the listener without bounce-paths, but suitably delayed to arrive synchronously with any (steered and bounced) higher harmonics."
I'm wondering what frequencies he's talking about, <80Hz, <500Hz, < 1000Hz? Is this low frequency delay even necessary despite the larger integration time of our hearing at those frequencies?
If such a LF delay is not necessary then the design could look similar to Stereolith but working differently. The reflections of Stereolith aren't properly aimed and too low in level. The front mounted tweeter takes precedence over the reflections created by the side firing drivers. As a result all sounds come from the center.
I'm looking for very loud reflections (side wall and 2nd order ceiling/side wall) while maintaining flat frequency response from the center.
Hooley also notes the breakdown of precedence that I've described some pages ago: "... we have determined that there is a second effect essentially the inverse of the Haas/ Precedence Effects; viz., if the later arriving (within ~10– 35 ms) sound copy is louder by 12 dB to 15 dB or more than the earlier arriving sound, then the brain completely ignores the early arriving sound, and hears, and direction- alises, only the later copy."
This is from a Steve Guttenberg review of one of the Yamaha sound bars. This is another of the DSP units that bounces multiple narrow beams of sound to create multple channel location. Summary: spacious but narrow separation.
"The YSP-3000, like all of the other Yamaha YSP speakers we've tested to date, is capable of projecting a large, immersive soundfield. No other brand's single-speaker surround systems are even close; Yamaha YSP speakers are state of the art in that regard. That holds true for listeners seated directly in front of the YSP-3000 and listeners seated off to the sides of the couch. We could even stand up and walk around the room and the sound remained spacious. However, while the speaker can be used without a subwoofer, we wouldn't recommend it as the YSP-300 won't produce any deep bass on its own. We used a Yamaha YST-FSW100 sub for most of our listening tests.
The Pirates of the Caribbean: At World's End Blu-ray disc sounded good enough for us to forget that all of the sound filling the CNET listening room was coming from the YSP-3000. Sound quality was generally quite good, but when the demands of the naval battle scenes upped the ante, the Yamaha speaker's limitations became obvious. The ships' cannon blasts didn't have the same impact we'd get from a 5.1 channel speaker/subwoofer combination like Aperion's Intimus 5B Harmony SD. The Yamaha YST-FSW100 subwoofer was partly to blame, as it didn't have the sock of the Aperion's Bravus 8D sub. The YSP speaker can also be overtaxed and sound strained by highly dynamic special effects. So overall surround sound was excellent, but judged on a sound-quality basis, the YSP speaker sounded less clear and detailed than Polk's SurroundBar 50 single-speaker surround system.
We switched to music, and the Talking Heads' Stop Making Sense concert DVD rocked pretty hard; the audience's cheers and claps were projected well forward of the speaker. CD sound was less satisfying overall, though it was definitely at its best when we listened in stereo. True, separation was fairly narrow in stereo mode, but engaging the YSP-3000's surround processing to create a more room-filling sound also adversely affected the sound quality, making it hazy, harsher, and generally less satisfying. In any case, the Black Key's blues-rock lacked punch and orchestral classical music had a muffled quality. The YSP-3000 was more enjoyable with Blu-ray and DVDs."
So highly spacious but not well defined makes it sound like another high diffusion approach.
On the Bose 901/Geddes continuum it is towards the left end.
David S.
"The YSP-3000, like all of the other Yamaha YSP speakers we've tested to date, is capable of projecting a large, immersive soundfield. No other brand's single-speaker surround systems are even close; Yamaha YSP speakers are state of the art in that regard. That holds true for listeners seated directly in front of the YSP-3000 and listeners seated off to the sides of the couch. We could even stand up and walk around the room and the sound remained spacious. However, while the speaker can be used without a subwoofer, we wouldn't recommend it as the YSP-300 won't produce any deep bass on its own. We used a Yamaha YST-FSW100 sub for most of our listening tests.
The Pirates of the Caribbean: At World's End Blu-ray disc sounded good enough for us to forget that all of the sound filling the CNET listening room was coming from the YSP-3000. Sound quality was generally quite good, but when the demands of the naval battle scenes upped the ante, the Yamaha speaker's limitations became obvious. The ships' cannon blasts didn't have the same impact we'd get from a 5.1 channel speaker/subwoofer combination like Aperion's Intimus 5B Harmony SD. The Yamaha YST-FSW100 subwoofer was partly to blame, as it didn't have the sock of the Aperion's Bravus 8D sub. The YSP speaker can also be overtaxed and sound strained by highly dynamic special effects. So overall surround sound was excellent, but judged on a sound-quality basis, the YSP speaker sounded less clear and detailed than Polk's SurroundBar 50 single-speaker surround system.
We switched to music, and the Talking Heads' Stop Making Sense concert DVD rocked pretty hard; the audience's cheers and claps were projected well forward of the speaker. CD sound was less satisfying overall, though it was definitely at its best when we listened in stereo. True, separation was fairly narrow in stereo mode, but engaging the YSP-3000's surround processing to create a more room-filling sound also adversely affected the sound quality, making it hazy, harsher, and generally less satisfying. In any case, the Black Key's blues-rock lacked punch and orchestral classical music had a muffled quality. The YSP-3000 was more enjoyable with Blu-ray and DVDs."
So highly spacious but not well defined makes it sound like another high diffusion approach.
On the Bose 901/Geddes continuum it is towards the left end.
David S.
Last edited:
guys we are drifting off topic again. I'm not sure whether we posted the link to the other thread we created when we first split off the more esoteric stuff. http://www.diyaudio.com/forums/mult...ques-achieving-oustanding-stereo-imaging.html
Tony.
Tony, that will probably kill this thread in a minute and I still don't see why a different thread is needed. It basically prohibits a meaningful discussion of how an optimal reproduction setup should look like - but you're the boss. Would you be so kind to move the last few posts about "stereo center" to the other thread although I don't think it belongs there.
Hi Markus, I certainly don't want to kill the thread (or the discussion)! I note that there hadn't been any discussion in the other thread since we split it and thought it might be off peoples radars. Pano is a bit more in tune with this so I'll wait till he weighs in (ie quite a bit of it is over my head )
Tony.
(ie quite a bit of it is over my head ) Tony.
Last edited:
Despite the title of the thread, I didn't think multichannel (more than 2 speakers) had been excluded from the discussion ?
Although full discrete 5.1 channel sound as is normal on DVD's is probably not directly relevant, and something of a different experience more optimized for movies, using additional speakers such as a derived centre or additional "wide" speakers for playback of stereo recordings to enhance the stereo experience, and the kinds of directivity those speakers would want does seem relevant.
It could be that 3 or 4 speakers may in the future be recognised as the optimal way to enjoy 2 channel material, once the best way to derive those extra channels is figured out...
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.