[dprimary]

Not exactly 20log(2/0.5) + AIr Absorption at 200 Hz ( neglectable) -> roughly 12dB vs 14dB.
( 200 Hz has a 1.5 meter wavelength, in principle the 1/r attenuation is not applicable interactions between waves are more complex. I did the above calculation just for copping with your example )

Is there a decimal missing?

 

[bigshot]

since we are comparing normal audio to ultrasonics, (the argument being... ultrasonics don't matter because they're too attenuated relative to normal sound to matter) isn't the relative atmospheric absorption the most relevant number?

I would think that the fact that ultrasonics are inaudible would be the most relevant, but people continue to discuss it anyway.

 

[Zapp_Fan]

Not exactly 20log(2/0.5) + AIr Absorption at 200 Hz ( neglectable) -> roughly 12dB vs 14dB.
( 200 Hz has a 1.5 meter wavelength, in principle the 1/r attenuation is not applicable interactions between waves are more complex. I did the above calculation just for copping with your example )

Yes, but roughly speaking, if something were audible at -12dB it'd still be audible at -14dB. Anyway, I am NOT saying ultrasonics are audible, but if they did have some value in the recording , like _________________ (this space left intentionally blank)... then we can't completely discount that possibility on the sole basis of attenuation between source and mic, relative to normal audio.

To put it another way, ultrasonics that originate at the same SPL as another frequency will typically only be a few dB lower at the mic position, at least speaking strictly about direct-path audio.

On the other hand, in some spaces most of the sound reaching a given point is NOT direct and is reflected, so in practice I'd (take a wild) guess you'd capture ultrasonics at no better than an -18dB penalty relative to 1Khz or something, because of absorption by stuff in the room... like walls. To further waver on the point, that doesn't mean you won't get something intelligible at the mic, just means it will be quieter than stuff that got there by bouncing.

 

[KeithEmo]

That's EXACTLY what we're talking abut here....
(I think some people came into the conversation late.)

I think many people here are confusing absorption with dropoff due to distance.
Of course, as you get further from the source, the sound level will get lower.
(And the math whereby that occurs will depend on the space the sound is radiated into.)
However, assuming that the absorption of all frequencies by the air was equal....
As we got further away the overall amplitude would be lower..... but it would still remain proportionally equal at all frequencies.
So, if the cymbals were at a certain SPL level at one inch, and 5% of their energy was at 30 kHz......
The level would somewhat lower at two meters, but 5% of the energy at that distance would still be at 30 kHz.....

HOWEVER, the reality is that the absorption of different frequencies by air is NOT at all equal.
The absorption by air at 200 Hz is very close to 0 dB.
The absorption at 30 kHz (under the other conditions I listed) is about 0.9 dB/meter.
So, as the distance increases, you will lose proportionally more amplitude as the frequency goes up.

So, for whatever distance you choose, you calculate the predicted amplitude, based on the distance dropoff calculations.
Then, for relatively high frequencies, you ADD the additional attenuation that occurs because of absorption by the air.
So, for example, at two meters, the level at 200 Hz will be what you would expect based on the "dropoff calculations"...
But, at two meters, the level at 30 kHz will be 3.6 dB lower than that due to the ADDITIONAL loss of 3.6dB due the energy absorbed by the air at that distance and frequency.

@gregorio's original assertion was essentially that:
"Even if there is a significant amount of 30 kHz being produced by the cymbal it won't matter because the air will absorb it all before it reaches the microphone."
He then went on to assert that the loss would be something like 50 dB at 20 meters.

However, those numbers are not correct.
According to the calculations, at two meters, which is what I specified, the energy at 30 kHz will have lost an additional 3.6 dB due to air losses in addition to the dropoff caused by the distance.
In other words, if you measure it, at two meters the cymbals should show an additional roll off of 3.6 dB at 30 kHz compared to the level at relatively low frequencies.
This is quite different than claiming that "nothing at 30 kHz would remain".
(Note that all of the numbers I quoted were for 50% relative humidity and 20 degrees Centigrade at sea level pressure).

We all realize that OVERALL level will drop with distance.
The discussion revolves around how much ADDITIONAL loss there will be at ultrasonic frequencies due to air absorption.
So, for example, if the total attenuation at 2 meters at 200 Hz calculates out to 14 dB, we would expect the drop off at 30 kHz to measure 17.6 dB (which is pretty far from "nothing measurable or recordable remaining").

OK, I get that, makes total sense. But then since we are comparing normal audio to ultrasonics, (the argument being... ultrasonics don't matter because they're too attenuated relative to normal sound to matter) isn't the relative atmospheric absorption the most relevant number?

I mean, yes, certainly, 14db total attenuation at 2m, but unless I'm missing something, that value would be similar at (say) 200hz also.

 

[KeithEmo]

But, then, that's another discussion altogether.

All we were ascertaining here was that we would actually be able to produce an appropriate test signal to determine whether that was true or not.
And, in order to do that, we must first produce a recording of cymbals with a frequency response that truly extends to 30 kHz.
Then we must find a pair of headphones or speakers that can reproduce it.
Once we have those we actually have the materials necessary to perform some serious audibility tests.

AT THAT POINT, we can test whether anybody can hear it, or whether they notice when we remove it.

This whole discussion is really a sort of detour associated with @gregorio's assertion that we would have difficulty creating the required test signal.
(Or something like that.)

I would think that the fact that ultrasonics are inaudible would be the most relevant, but people continue to discuss it anyway.

 

[KeithEmo]

I absolutely agree.....

I would expect surface absorption at ultrasonic frequencies to be quite high...
Although, since most rooms probably aren't designed with much consideration given to that, we might all be surprised there in some cases.

I should also point out that, in terms of audible frequencies, our brains seem able to differentiate first-arrival direct sounds from later arriving reflected sounds.
And our brains interpret the difference as "room acoustics".
(Even though a bright speaker in a heavily padded room, and a dull speaker in a live room, may have the same long term average response, they still sound different to we humans.)
There is also the possibility, which some studies seem to suggest, that our brains actually can detect fine differences in wave forms and arrival times...
Some studies also seem to suggest that the limits of audibility for this may be different than those for pure continuous sine wave test tones.
A few studies even suggest that we are able to detect differences in arrival time of as little as a few microseconds under some conditions.

Please not that I am NOT specifically claiming that we would in fact experience cymbals as sounding different depending on whether our recording includes frequencies up to 30 khz or so.
However, since the results of some studies, as well as piles of anecdotal evidence, suggest that this might be the case...
I'm simply suggesting that it might be nice to actually do some testing to find out - rather than continuing to accept the results of often outdated and flawed tests as "absolutely reliable facts".

Yes, but roughly speaking, if something were audible at -12dB it'd still be audible at -14dB. Anyway, I am NOT saying ultrasonics are audible, but if they did have some value in the recording , like _________________ (this space left intentionally blank)... then we can't completely discount that possibility on the sole basis of attenuation between source and mic, relative to normal audio.

To put it another way, ultrasonics that originate at the same SPL as another frequency will typically only be a few dB lower at the mic position, at least speaking strictly about direct-path audio.

On the other hand, in some spaces most of the sound reaching a given point is NOT direct and is reflected, so in practice I'd (take a wild) guess you'd capture ultrasonics at no better than an -18dB penalty relative to 1Khz or something, because of absorption by stuff in the room... like walls. To further waver on the point, that doesn't mean you won't get something intelligible at the mic, just means it will be quieter than stuff that got there by bouncing.

 

[Arpiben]

Even at small Jazz club my ears are rarely less than 5 meters from cymbals and almost never in free line of sight (no obstacles).
A few posts back I did provide @KeithEmo with closed mic ed cymbals hits. At that time I was requesting how he could retrieve room characteristics from its ultrasonic content.
We have files available.
Who is turning in circles?

 

[KeithEmo]

I have to end my posting today with a funny story (I don't know if it qualifies as a fable).

As background, there is an old "wives tale" that "you can capture a bird by sprinkling salt on his tail".
(I have no idea where it came from, but it's been around for a very long time, and seems logically to be quite silly.)

HOWEVER, an old Sylvester and Tweety cartoon provided an excellent lesson in how our biases can affect how we interpret things.
If you didn't know, Sylveser is a talking cat, and his lifetime obsession is capturing and eating Tweety, who is a talking canary bird.

In one cartoon, Sylvester approaches Tweety, who has escaped his cage and is sitting on a window ledge.
As usual, Sylvester is ready to pounce, but Tweety is ready to fly, and he is faster than the cat.
Sylvester approaches cautiously, and says to Tweety: "I read somewhere that I can keep you from escaping by sprinkling salt on your tail".
Tweety of course replies: "That's silly, that's just an old wives tale."
About two seconds later the five pound sack of salt lands on top of Tweety.

I believe the moral of the story is that SOMETIMES, old wives tales, and anecdotal evidence in general, actually does have some basis in fact.

 

[KeithEmo]

Indeed.
And I do deserve you a response.
I personally was never able to retrieve any useful information regarding room acoustics from those files.
However, that doesn't prove that it's impossible, but merely that I was unable to do so.... and those are very different things.
For example, I could draw a flowchart describing one method which I think might be able to do so...
However, I have neither the DSP programming skills to write the code myself, nor the inclination to pay what it would cost to have someone else write it...
(And, yes, that particular idea may turn out to not be successful.)

Far too many people in this thread seem unable to tell the difference between "hasn't been done yet that I know of" and "impossible"

I should also stress the difference between "detailed color photos" and "information".
Not all information need be complete or even useful.

And, if you accept that, I actually can provide you with an obvious example.
Assuming we have a spot where a cymbal is struck after a few seconds of relative silence.
(Lack of any significant high frequency content for a few seconds will do.)
We should be able to pick out the envelope shape of the initial cymbal strike.
Then, starting from there, if we see an envelope of similar shape 20 mSec later...
We can now surmise that there is some surface whose "bounce time" is 20 mSec...
So we will know that the first reflection traveled over a 20 foot signal path...
And, if we compare the spectra of the two, we will learn something.
We will see a difference in spectra due to air and surface absorption at various frequencies.
When we correlate this with other information....
We may be able to determine what the nearest wall is covered with...
Or what the humidity was when the recording was made...
It's sort of like an equation with many variables...
We may determine that EITHER the wall was covered with felt OR it was damp that day.
Then, by combining that with other information, we may be able to tell which applies this time.
This is not at all like having a program present us with a 3D image of the room.
(But, then, if you've seen original RADAR images, they don't look like much either.)

Even at small Jazz club my ears are rarely less than 5 meters from cymbals and almost never in free line of sight (no obstacles).
A few posts back I did provide @KeithEmo with closed mic ed cymbals hits. At that time I was requesting how he could retrieve room characteristics from its ultrasonic content.
We have files available.
Who is turning in circles?

 

[bfreedma]

Indeed.
And I do deserve you a response.
I personally was never able to retrieve any useful information regarding room acoustics from those files.
However, that doesn't prove that it's impossible, but merely that I was unable to do so.... and those are very different things.
For example, I could draw a flowchart describing one method which I think might be able to do so...
However, I have neither the DSP programming skills to write the code myself, nor the inclination to pay what it would cost to have someone else write it...
(And, yes, that particular idea may turn out to not be successful.)

Far too many people in this thread seem unable to tell the difference between "hasn't been done yet that I know of" and "impossible"

A bigger problem is people equating “it hasn’t been done yet” to “when it’s done, it will be audible to humans”

 

[Zapp_Fan]

I absolutely agree.....

I would expect surface absorption at ultrasonic frequencies to be quite high...
Although, since most rooms probably aren't designed with much consideration given to that, we might all be surprised there in some cases.

I should also point out that, in terms of audible frequencies, our brains seem able to differentiate first-arrival direct sounds from later arriving reflected sounds.
And our brains interpret the difference as "room acoustics".
(Even though a bright speaker in a heavily padded room, and a dull speaker in a live room, may have the same long term average response, they still sound different to we humans.)
There is also the possibility, which some studies seem to suggest, that our brains actually can detect fine differences in wave forms and arrival times...
Some studies also seem to suggest that the limits of audibility for this may be different than those for pure continuous sine wave test tones.
A few studies even suggest that we are able to detect differences in arrival time of as little as a few microseconds under some conditions.

Please not that I am NOT specifically claiming that we would in fact experience cymbals as sounding different depending on whether our recording includes frequencies up to 30 khz or so.
However, since the results of some studies, as well as piles of anecdotal evidence, suggest that this might be the case...
I'm simply suggesting that it might be nice to actually do some testing to find out - rather than continuing to accept the results of often outdated and flawed tests as "absolutely reliable facts".

Hmm, this suggests a way to guess whether such a test is plausible without having to get out of our chairs and build this rig.

Do we know any SPL thresholds or similar data from those experiments that could be generalized to other audio? It should be easy enough to estimate the SPL of ultrasonics in a normal listening setup, based on existing recordings, and go from there

 

[analogsurviver]

Even at small Jazz club my ears are rarely less than 5 meters from cymbals and almost never in free line of sight (no obstacles).
A few posts back I did provide @KeithEmo with closed mic ed cymbals hits. At that time I was requesting how he could retrieve room characteristics from its ultrasonic content.
We have files available.
Who is turning in circles?

Just yesterday, I recorded a binaural concert of jazz - by young musicians - in the legendary venue

https://www.prulcek.si/

with cymbals less than 5 m from my position, cursing the note stand the leader of the band, a trumpet player, decided to put exactly in "line of fire". And I had a trumpet ( albeit not muted - muted trumpet is specially rich in ultrasonics, beyond 80 khz ) and sax on less than 3 metre .

As the gig has been partially amplified ( electric guitar, double bass, most - but not all - of the time trumpet ), I chose to record only in DSD64 - which can be problematic regarding inherent DSD ultraosnic noise. Still, a pretty good picture of what is going on up to at least 48 kHz ( 96kHz sampling in PCM for spectrum analysis ) should be possible.

 

[Arpiben]

Indeed.
And I do deserve you a response.
I personally was never able to retrieve any useful information regarding room acoustics from those files.
However, that doesn't prove that it's impossible, but merely that I was unable to do so.... and those are very different things.
For example, I could draw a flowchart describing one method which I think might be able to do so...
However, I have neither the DSP programming skills to write the code myself, nor the inclination to pay what it would cost to have someone else write it...
(And, yes, that particular idea may turn out to not be successful.)

Far too many people in this thread seem unable to tell the difference between "hasn't been done yet that I know of" and "impossible"

I should also stress the difference between "detailed color photos" and "information".
Not all information need be complete or even useful.

And, if you accept that, I actually can provide you with an obvious example.
Assuming we have a spot where a cymbal is struck after a few seconds of relative silence.
(Lack of any significant high frequency content for a few seconds will do.)
We should be able to pick out the envelope shape of the initial cymbal strike.
Then, starting from there, if we see an envelope of similar shape 20 mSec later...
We can now surmise that there is some surface whose "bounce time" is 20 mSec...
So we will know that the first reflection traveled over a 20 foot signal path...
And, if we compare the spectra of the two, we will learn something.
We will see a difference in spectra due to air and surface absorption at various frequencies.
When we correlate this with other information....
We may be able to determine what the nearest wall is covered with...
Or what the humidity was when the recording was made...
It's sort of like an equation with many variables...
We may determine that EITHER the wall was covered with felt OR it was damp that day.
Then, by combining that with other information, we may be able to tell which applies this time.
This is not at all like having a program present us with a 3D image of the room.
(But, then, if you've seen original RADAR images, they don't look like much either.)

Indeed.
And I do deserve you a response.
I personally was never able to retrieve any useful information regarding room acoustics from those files.
However, that doesn't prove that it's impossible, but merely that I was unable to do so.... and those are very different things.
For example, I could draw a flowchart describing one method which I think might be able to do so...
However, I have neither the DSP programming skills to write the code myself, nor the inclination to pay what it would cost to have someone else write it...
(And, yes, that particular idea may turn out to not be successful.)

Far too many people in this thread seem unable to tell the difference between "hasn't been done yet that I know of" and "impossible"

For your purpose I am afraid you will need much more ultrasonic power than the one available in music recording.Not even talking about having the mics characteristics and exact positioning during the event....
If not already done you may check power involved in ultrasonic applications and their distance range.

 

[KeithEmo]

That is an interesting and dangerous question... the danger, as usual, arises from the fact we always end up making assumptions about how far we can generalize things, so we run the risk of being wrong. And, when it comes to audio science, sometimes you just can't know for sure whether the proverbial rabbit hole has an exit or not unless you reach the end of the tunnel.

From the studies I've been able to find, it is widely agreed that we hear tones when the nerve fibers in our ears associated with a given frequency are stimulated. Like little tuning forks, each fiber corresponds to a different frequency; when a given fiber is stimulated, we "hear" that frequency. And it is pretty widely agreed that the range of "tuning forks" in the human ear respond to frequencies between 20 Hz and 20 kHz. This is the basis for the claim that "our range of hearing extends from 20 Hz to 20 kHz", which has been pretty widely tested (although there have been slight differences in results).

However, there seem to be other nerve fibers in there that respond to other things, and that send signals when specific things happen, like "when a sound starts". And these haven't been researched in nearly as much detail. And, just to make things more complicated, the output of all of these nerves is processed by our brains, in all sorts of interesting ways. And then we get into things like masking effects, which derive from how our brain does all that processing.

This brings up all sorts of interesting questions which really have NOT bee answered in detail.

Here's an easy but interesting idea to start...

We all know that loud sounds at a certain frequency mask our ability to hear quieter sounds at nearby frequencies. The louder the sound the more effectively it masks quieter sounds at nearby frequencies and prevents us from hearing them. Just to make life interesting, the effect varies depending on the frequency of the loud tone, the distance between it and the quieter tones, and whether they are above or below it. I believe it also varies depending on the absolute loudness of both tones... and it also depends on what else you're hearing at the time.

So, here's my interesting question....

Let's assume I play a relatively quiet level of steady white noise.

If I then switch a loud 22 kHz tone on and off will it cause an audible change, not because I hear it, but because it is applying a masking effect to frequencies I can hear. (Will my white noise sound duller when the ultrasonic tone is playing because it's masking some of the upper audible frequencies?)

This sounds like a relatively simple test....
And also one whose results would be quite interesting.....
Yet I have never been able to find any published results of anyone having actually done it.

From what I can tell, particularly when it comes to audio, we are all operating under a worrisome number of assumptions, and very little real test data.

Hmm, this suggests a way to guess whether such a test is plausible without having to get out of our chairs and build this rig.

Do we know any SPL thresholds or similar data from those experiments that could be generalized to other audio? It should be easy enough to estimate the SPL of ultrasonics in a normal listening setup, based on existing recordings, and go from there

 

[bigshot]

A bigger problem is people equating “it hasn’t been done yet” to “when it’s done, it will be audible to humans”

One very good reason that a thing might not have been done yet is because it might not need to be done.

I'm trying to imagine how loud an ultrasonic frequency would have to be to reach from the stage of the Hollywood Bowl to the benches along the back wall of the audience. I bet the ears and brains of everyone in the front two thirds of the audience would melt!