Here is a hypothesis that could help-
If we accept and understand that the pathway from the acoustic space in front of an ear to the cochlea is non-linear then it can function to "demodulate" a signal that is not directly audible. The pathway will have substantial attenuation to higher frequencies but not infinite.
Then it may be possible to generate ultrasonic tonebursts that would be audible after the "demodulation" of the acoustic path.
This should be easy enough to test. I can give it a quick try tomorrow and if something comes of it then others can see where I overlooked the obvious.
My understanding is that at higher frequencies the pitch resolution is not precise, one of the aspects that makes lossy compression work.
As for time resolution you do not need frequency response to the reciprocal of the time to resolve. A time interval analyzer just needs to identify the threshold crossing somehow. My TIA (HP5370A) with 100 MHz bandwidth and 20 pS one shot resolution are a verification of this basic fact.
Hans, there's nothing wrong with tests beeing 20+ years old. One has to remember that all the pioneering work was done back then, when all of it was new and exciting. I haven't seen that kind of collaboration between different parties involved in HD developement ever since. I was (and still am) on that mailing list mentioned in the article and witnessed everything first hand.
I am sure, there are similar tests being conducted even today, but, I suspect, they are now oriented more towards making profit. Everyone is tooting his own horn now - 192k this, 384k that, DSD, double DSD, more is better, bla bla bla. There are a few sensible voices out there, but they are all lost in the overall noise produced by internet "experts", opinion makers an five dollar chip repackagers.
IIRC, there were a few documents referenced in the paper by V.Melchior I mentioned the other day, but they all are AES papers and must be purchased. I'll see what else similar I can dig up, but no guarantee..
I myself have been a firm believer in that "it's in the filters" (all else being equal) for the past 20 years. Worked for me, to quote JC.
Last edited:
180 degrees unmoot. You want an enclosure to be as small as possible so that it does not support standing waves at the generated frequencies. Anything that makes the enclosure looks larger to sound waves is bad.
When a woofer box reaches 'sufficient size', you start having standing wave problems, so it is a remarkable choice of words you used.
elektroj - "I myself have been a firm believer in that "it's in the filters" (all else being equal) for the past 20 years."
Of course it's in the filters, and it's all one filter after another from pickup to loudspeaker (which of course is the worst filter of them all).
I can recall reading a discussion on why a brickwall filter is less savory than one with a kinder, gentler slope 20 years ago. 20+ years later, still talking about it. We knew it was in the filter, when 192k was just someone's dream.
I would think they'd have it all worked out by now, perhaps different filters for each of the different sample rates. Rather then just run 'em all through the lowest frequency filter...
Of course it's in the filters, and it's all one filter after another from pickup to loudspeaker (which of course is the worst filter of them all).
I can recall reading a discussion on why a brickwall filter is less savory than one with a kinder, gentler slope 20 years ago. 20+ years later, still talking about it. We knew it was in the filter, when 192k was just someone's dream.
I would think they'd have it all worked out by now, perhaps different filters for each of the different sample rates. Rather then just run 'em all through the lowest frequency filter...
You can certainly choose that route. It means low efficiency if you want to go low. So big amps.
Some would rather have high efficiency and low bass which means BIG boxes. There are also big boxes that use resonance to advantage (TLs, Horns).
Choose your poison. Most of my speaker designs fall closer to the former than the latter. The alignment i use tends to smallish boxes. Still some of the big, efficient drivers need fairly big boxes.
Here is a HiEff woofer box visualization. You can see the bracing acting as a device to break up standing waves (look at how the holes line up).
dave
I was talking about the filters in digital signal processing, not filters in general.
IMO, Bottom line --- the problem with filters persists because the sharp cut-off filters are near 20Khz and they need to be further away at 40+ KHz.
We need a new standard.
THx-RNMarsh
Last edited:
As long as the wavelength inside the cabinet is much longer than cab size AND/OR the cross-over is below that cabinet dimension related freq... no issues with standing waves occures.
-RNM
Last edited:
reporting a post is enough. No need to also post about it in the thread. One post deleted as collateral.
Pitch perception is an interesting and complicated phenomena, there are a number of aural tricks that can be played.
Of course, though this audiophile myth perseveres mainly due to repeated propagation by those in the audio business
unless they really don't understand the basics, they could still learn.....
Do you think they haven't? Different filter options, when available, are for audiophiles to fiddle about with
> Commercial digital filters in IC form usually scale with the sampling frequency.
> That is the most cost effective way. And most logical, IMO.
> There might be a different and switchable set of coeficients included in ROM, but that's about it.
But that's ideal. Store some extra coefficient sets in ROM and have pull ups & unpopulated zero-ohm bridges to address them. Jumpers are too easy. Like the op amp roller, the filter conoisseur wants the feeling that he's seriously working on the leading edge of technology.
Leak the information that the other filters are reserved for a mythical high end model and they were so costly to implement that you cannot possibly give them away, even on the same hardware. That gets them started for sure.
It has happened already. See the threads on EEVblog for unthrottling some Chinese scopes.
> That is the most cost effective way. And most logical, IMO.
> There might be a different and switchable set of coeficients included in ROM, but that's about it.
But that's ideal. Store some extra coefficient sets in ROM and have pull ups & unpopulated zero-ohm bridges to address them. Jumpers are too easy. Like the op amp roller, the filter conoisseur wants the feeling that he's seriously working on the leading edge of technology.
Leak the information that the other filters are reserved for a mythical high end model and they were so costly to implement that you cannot possibly give them away, even on the same hardware. That gets them started for sure.
It has happened already. See the threads on EEVblog for unthrottling some Chinese scopes.
I've not figured out what audiophile myth you are referring to. Would you please explain?Of course, though this audiophile myth perseveres mainly due to repeated propagation by those in the audio business
Jn
I didn't know about the skin route.
But 40 years later, I still remember the smell. That and xylene. And polyurethane, after smelling that, I can taste it for a day or so.
Somebody told me acetone when smelled, can be detected in the liver 15 seconds later.
Jn
Go up-town! Use DIP-14 sockets and removable pin headers, programmed by scramble-wiring. If you program it using only two wires, you get (14 choose 2) * (12 choose 2) different possibilities for someone to try. That's 91 * 66 = 6006 possible programmings. But it chews up 14 I/O pins on your uC.
With DIP-8 and 3 programming wires you get (8 C 2) * (6 C 2) * (4 C 2) = 28 * 15 * 6 = 2520 unique programmings, and it "only" takes 8 I/O pins on your uC.
Feel free to double check my calculation of the binomial coefficient (n choose k), here are some online resources
Free Binomial Coefficient Calculator - Free Statistics Calculators
Binomial Coefficient Calculator
Binomial Coefficient Calculator