OK, you lost me there. I just saw some nice vertical and horizontal plots of various speakers all done on the same rig. "Hey, nice." But why not include all 360º so we have a full picture? That would be a nice bit of info.
.
Sorry, thought you were commenting on my curves of center channel tonal error, posted in excel.
Ben, what is helpful is to remember the stereo illusion is an illusion and it's much improved if there is cross talk cancellation. And the better the cancellation is performed, to that degree the stereo illusion is more convincing.
Instead of dismissing out of hand, (as you have done), the fellow's unfortunate background as an extremely successful engineer and academic, it's better to think of the possibility he's developed a better cross talk cancellation algorithm to use with speakers.
What's your problem?
He got a grant from a Princeton foundation that is described by it's benefactor as a "slush fund" for tinkerers, and she thinks the folk at Princeton "should lighten up." Good for her, I say. She's getting her money's worth.
Grants in Princeton alum's name allow professors, students to pursue off-beat passions | NJ.com
I have a feeling the folk in TO should lighten up, as well.
My point in mentioning the bibliography is that the guy is going about his tinkering the right way.
Your parallel with bridge building is not a good one. Either the algorithm gives results that measure better than previous ones, or it doesn't. Human testing is in this case relatively easy as there is only one variable - the algorithm;
Two if you change speakers
I think I know what his motivation is. Looking at his biography it's clear he doesn't have to worry about getting grants, but he does have @7000 recordings in his personal library....😀
Instead of dismissing out of hand, (as you have done), the fellow's unfortunate background as an extremely successful engineer and academic, it's better to think of the possibility he's developed a better cross talk cancellation algorithm to use with speakers.
What's your problem?
He got a grant from a Princeton foundation that is described by it's benefactor as a "slush fund" for tinkerers, and she thinks the folk at Princeton "should lighten up." Good for her, I say. She's getting her money's worth.
Grants in Princeton alum's name allow professors, students to pursue off-beat passions | NJ.com
I have a feeling the folk in TO should lighten up, as well.
My point in mentioning the bibliography is that the guy is going about his tinkering the right way.
Your parallel with bridge building is not a good one. Either the algorithm gives results that measure better than previous ones, or it doesn't. Human testing is in this case relatively easy as there is only one variable - the algorithm;
Two if you change speakers
I think I know what his motivation is. Looking at his biography it's clear he doesn't have to worry about getting grants, but he does have @7000 recordings in his personal library....😀
DDF, I know you've read this on other forums, but how do you reconcile your theory with the fact that some one has mixed and mastered these recordings? That should account for HRTF and is thus stopped dead in it's tracks right there. Also, not everyone has this idealized🙁 equilateral triangle? Ignoring the room issues of course.
Thanks,
Dan
Thanks,
Dan
Hi Mike,
Crossfeed is to moderate that annoying "sound in the middle of the head" effect that is often mentioned as a key factor for people's dislike for headphones. The crossfeed, as I understand it, is intended to provide positional and spatial cues that indicate direction, proximity and perhaps even the ambiance that are experienced naturally. Even Binaural recordings, which are perhaps the most accurate recording that can be played into headphones, depend on a certain amount of natural crossfeed in each channel, just as we experience while listening in real life.
Crosstalk, in the discussion of audio devices, is when there are pure signals leaking into the other channel, when they shouldn't. At least this is what I've been led to believe.
Best Regards,
TerryO
Dan, I had answered that in my first post: Standard pan pots don't eq this out. I have never heard of anyone that accounts for this on the recording side, so I don't believe it is common to do so at all. If you have evidence showing otherwise, I'd be more than happy to see it.
Also, it isn't eq'ed out for minimalist (two) mic'ed recordings, like you'll find with some classical music.
I also said in my first post that it wasn't applicable as shown to everyone due to the uniqueness of the HRTF but represents an average and I said in my first post "room notwithstanding". That was all in my original post, which was rather short so should be hard to miss.
I hope this doesn't turn into one of those dreadful threads from the other forum, I left there for a reason. Please read my original post before critiquing it.
This is an error in stereo as I point out. The fact that other errors exist doesn't invalidate the existence of this error. If its not a perfect equilateral triangle, then the shape of the error curve will change, but not by much unless you decide to sit way out of the standard stereo set up.
Dave
Problem with headphone crosstalk cancellation is that the response must be some pre-canned curve. When applied to your unique head response, there will be some tonal distortion. For some people, it will be alot.
Secondly, it can't create images on the "cone of confusion", zones in space where the spatial cues aren't unique to that locatiuon. This requires head motion. Sony marketed a headphone with an accelerometer that allowed head movement, to try and remove the ambiguity on the cone of confusion.
We bought one, tested it, for me the illussion didn't work.
Dave
Stop getting emotional Dave and there won't be any problems. No one said standard pan pots EQ anything. I could just as easily say read my post at this point. The fact that people are listening the recording to make it sound good is common knowledge. Find an engineer who isn't doing that and I'll gladly give some credence to your theory. How well they accomplish this is up to the up to the end user to decide, but I have flat measuring speakers with great off axis performance and suddenly I have more great recordings. Ignoring Dr. Tools, Olive, Geddes, etc... And defending contrary theories with only anecdotal evidence isn't my thing. I intend no offense by stating this. The various recording processes make designing a non standard ideal speaker an even dicier proposition. We need standardization, not more deviation.
Respectfully
Dan
Respectfully
Dan
Last edited:
Yep, everyone has 100% choice. I never understand why everyone can not just accept that everyone has their own perference and there isnt a truely right or wrong.
I like using Active XOs with full control over my EQing, I like to set a room curve that is tilted downward with a slight boost in the bass response 10Hz to 200Hz.
With my latest waveguide designs I even dipped the midrange a little because I found that I like less "glare" in that region.
In a perfect world Im sure all the great theories are the best appoarch but we do not live a perfect world so lets not worry about the theory of perfection. Its never been proven to be worth it to everyone anyways. Those who deem it needed should seek it.
btw, are your speakers flat in room or just measured flat then you allow for room gain?
Moving to interesting territory. Not sure I can follow this thought to the end, but I tried to think about areas where individual taste matters (no disputing that, as the Romans said in nice Latin), but other areas do have objective standards.
Some subject-matters, like wine tasting, are pretty soft and have no physical measurements. Sometimes there is expert judgment that is respected and sometimes we accept amateur judgment when it is long-term judgment or when longer perspective is added to initial "first taste" impressions (for sure with audio, great errors arise in testing when people are used to and familiar with a certain sound ("my mother's burnt soup" phenomenon). Are expectations rising due to auditioning better speakers (my standards changed last winter when I visited a friend with Martin Logans where you could hear the guitar squeaks across the hall).
There are even areas where majority opinion matters - it is a whole new area for web theorists now (I forget the cute name for it) - such as reading jumbles of letters that OCR can't decipher without human help.
Is it an "acoustic engineering" skill when the recording engineer cranks up an important flute passage during a Bruckner symphony... and everybody says, "Great recording"? Sure has nothing to do with cross-head-frequency-issues.
Sometimes hard to pin down what is changing over time as when dantheman finds his recordings are getting better! Mine too, as I add more tweeter-power over the years.
So where is music reproduction at home on those scales?
Some subject-matters, like wine tasting, are pretty soft and have no physical measurements. Sometimes there is expert judgment that is respected and sometimes we accept amateur judgment when it is long-term judgment or when longer perspective is added to initial "first taste" impressions (for sure with audio, great errors arise in testing when people are used to and familiar with a certain sound ("my mother's burnt soup" phenomenon). Are expectations rising due to auditioning better speakers (my standards changed last winter when I visited a friend with Martin Logans where you could hear the guitar squeaks across the hall).
There are even areas where majority opinion matters - it is a whole new area for web theorists now (I forget the cute name for it) - such as reading jumbles of letters that OCR can't decipher without human help.
Is it an "acoustic engineering" skill when the recording engineer cranks up an important flute passage during a Bruckner symphony... and everybody says, "Great recording"? Sure has nothing to do with cross-head-frequency-issues.
Sometimes hard to pin down what is changing over time as when dantheman finds his recordings are getting better! Mine too, as I add more tweeter-power over the years.
So where is music reproduction at home on those scales?
Last edited:
Doug, you've seen their measurements, they're flat psuedo anechoic and fall off in room. Nothing wrong with EQ, I just find as I work out acoustic problem I need it less and less, but they can help mask acoustic problems. They just won't fix them.
My recordings are the same, but my playback conditions have radically improved. Ben, what you are referring to is just the tip of the iceberg as I'm sure you are aware.
Dan
My recordings are the same, but my playback conditions have radically improved. Ben, what you are referring to is just the tip of the iceberg as I'm sure you are aware.
Dan
Yes, but the proper model is essential to settling some debated issues. It would be possible (but not useful) to advocate standardized recording and playback if tastes differ and music publishers couldn't make different strokes for different folks.
Listen to the studies of Toole and Olive. They say experts don't seem to agree well and almost every group does agree on what is best. Anybody on this forum like to hear that a whole bunch of high school kids and high-end audio salespeople are the brains that matter in shaping loudspeakers?
Their work is kind of experimentally clean but various "threats to validity" (as we say in the trade) can be present, esp. the "burnt soup" or adaptation level artifacts.
It looks like I missed the discussion on crosstalk and crossfeed. I'll still throw in a few cents worth.
I remember when the first Polk SDA series came out and Carver was first playing with his crosstalk cancelling system. Someone published an article on how to do what Polk was doing with 4 compact speakers. (Left pair, right pair, outter speakers with out of phase opposite channel?) We set it up in the lab at JBL and had great fun with it. "Money" from Dark Side of the Moon was the best demo. The jangling change was far left and far right, well outside of the speakers. A very neat effect. Of course you had to sit with your head in a vice.
The system worked by creating null zones that let the left ear hear the left speaker but cancelled out at the right ear's location (and vice versa). Seperation was dramatically increased. The electronic systems do the same thing. The two speakers on either side of a divider, with your nose to the divider, does the same thing.
When you have high separation then some recordings will sound unnatural. Anything pan potted hard left or right will sound unnatural if the opposite ear hears nothing. In nature we aren't used to that much seperation. The head related transfer function (HRTF) maxes out at about 20dB separation at low treble frequencies. The only way to get beyond that naturally is if someone whispers in one ear. The close proximity to one ear, and relatively far distance to the other (the ratio of distances is what matters) can increase the dB difference beyond the far field HRTF. At my current job we are experimenting with headrest speakers to give a sensation of near sound sources, hopefully different than the screen and surround speakers give.
As to crosstalk enhancement vs. crossfeed, they do seem to be working in opposite directions. Crossfeed was applied to headphones so that the separation wouldn't be too extreme. Done right you should be able to place sources with a standard pan-pot and yet get the right frequency response at each ear for all postions of panning. Headphone sound should be a little more natural that way, although it still probably won't come outside of the head (head motion tracking seems to be required for that).
I think the fundamental truth here is that the recording should match the playback system. Binaural goes well with enhanced separation schemes (including headphones). Standard speaker placements work well with panpotted mixes, even if hard panned full left or full right. A hard panned signal just moves to the location of that speaker. If it is 30 degrees left, then that is perfectly natural.
There also seems to be some room for personal preference here. I like music over headphones. It gives an expanded spread and I've gotten used sound in my head. Others can't stand headphones for that reason.
By the way, if you want to play back binaural over speakers it is just as easy to widen a stero pair out to 120 degrees or so and sit between them. It maximises the left right difference and is pretty close to headphone perspective.
David S.
I remember when the first Polk SDA series came out and Carver was first playing with his crosstalk cancelling system. Someone published an article on how to do what Polk was doing with 4 compact speakers. (Left pair, right pair, outter speakers with out of phase opposite channel?) We set it up in the lab at JBL and had great fun with it. "Money" from Dark Side of the Moon was the best demo. The jangling change was far left and far right, well outside of the speakers. A very neat effect. Of course you had to sit with your head in a vice.
The system worked by creating null zones that let the left ear hear the left speaker but cancelled out at the right ear's location (and vice versa). Seperation was dramatically increased. The electronic systems do the same thing. The two speakers on either side of a divider, with your nose to the divider, does the same thing.
When you have high separation then some recordings will sound unnatural. Anything pan potted hard left or right will sound unnatural if the opposite ear hears nothing. In nature we aren't used to that much seperation. The head related transfer function (HRTF) maxes out at about 20dB separation at low treble frequencies. The only way to get beyond that naturally is if someone whispers in one ear. The close proximity to one ear, and relatively far distance to the other (the ratio of distances is what matters) can increase the dB difference beyond the far field HRTF. At my current job we are experimenting with headrest speakers to give a sensation of near sound sources, hopefully different than the screen and surround speakers give.
As to crosstalk enhancement vs. crossfeed, they do seem to be working in opposite directions. Crossfeed was applied to headphones so that the separation wouldn't be too extreme. Done right you should be able to place sources with a standard pan-pot and yet get the right frequency response at each ear for all postions of panning. Headphone sound should be a little more natural that way, although it still probably won't come outside of the head (head motion tracking seems to be required for that).
I think the fundamental truth here is that the recording should match the playback system. Binaural goes well with enhanced separation schemes (including headphones). Standard speaker placements work well with panpotted mixes, even if hard panned full left or full right. A hard panned signal just moves to the location of that speaker. If it is 30 degrees left, then that is perfectly natural.
There also seems to be some room for personal preference here. I like music over headphones. It gives an expanded spread and I've gotten used sound in my head. Others can't stand headphones for that reason.
By the way, if you want to play back binaural over speakers it is just as easy to widen a stero pair out to 120 degrees or so and sit between them. It maximises the left right difference and is pretty close to headphone perspective.
David S.
Great post Dave! One thing about the binaural playback over speakers: shouldn't it require a pretty dead room and/or narrow dispersion speakers even when placed at 120 degrees to work well or will reflections not kill the effect d/t the first arrival's dominance/Haas effect? IOW, just the opposite of SL's recommendations.
Thanks,
Dan
Thanks,
Dan
Ignoring Toole? I was reading it 15 years ago when it came out.
Don't worry, this is the last post of yours I'll respond to. 1 week ago you didn't even know what HRTF was, and it shows. You don't understand the complexities of these issues.
I haven't thought about that. Whenever I try it, I leave my speakers in their usual position and just rotate them inwards and walk up in between them. I guess that gets me pretty close and increases the direct to reflected sound ratio.
I also find that picks up other attributes of headphones. Centered images are more in the head rather than in front. The other interesting effect is that the phasiness of speakers diminishes. That is, if you have a pair of speakers in the usual position and you shift left or right then, for any centered images there is a comb filtering effect from cancelation with unequal path lengths. If you move your speakers far to the sides then each ear hears less of the opposite speakers and the comb filtering diminishes or goes away. When you shift left to right the image smoothly swings without the "swish swish".
All off topic, but interesting.
David S.
Aha! Back to "flat"!
I mentioned the Bauer head-in-the-way concept a while ago. This link talks about it and includes the Bauer curve.
Stereophile: HeadRoom Supreme headphone amplifier
I built something similar, albeit passive with a big inductor in it, many decades ago. Very helpful for headphone listening.
As Dave S moves his speakers to different angles, the Bauer curve has different impacts and changes the tonal composition.
Of course, when you think about it, coloration changes whenever you move your head even considered ear by ear, let alone together separated by a head.
There is no inherent or necessary flat curve as much a math modelers wish there were - it just depends on a lot of things. So, added to "equal loudness" curves, we also have the Bauer issue.
What's that old audio salesman joke, "... yes sir, these headphones have rock solid separation."
I mentioned the Bauer head-in-the-way concept a while ago. This link talks about it and includes the Bauer curve.
Stereophile: HeadRoom Supreme headphone amplifier
I built something similar, albeit passive with a big inductor in it, many decades ago. Very helpful for headphone listening.
As Dave S moves his speakers to different angles, the Bauer curve has different impacts and changes the tonal composition.
Of course, when you think about it, coloration changes whenever you move your head even considered ear by ear, let alone together separated by a head.
There is no inherent or necessary flat curve as much a math modelers wish there were - it just depends on a lot of things. So, added to "equal loudness" curves, we also have the Bauer issue.
What's that old audio salesman joke, "... yes sir, these headphones have rock solid separation."
Last edited:
- Status
- Not open for further replies.