Home |
Search |
Today's Posts |
#1
|
|||
|
|||
Sampling Theory
Around the end of 2003, on a thread called 384KHz (regarding fast sampling),
I promised to post an artical on sampling with some focus regarding the ridiculess 384KHz and 192KHz trend. It took some work, and it is a long paper. If interested, go to www.lavryengineering.com Click on support, and it is the first artical under "white papers". It is called "Sampling Theory". It is a pdf artical and can be downloaded and printed, but the plots are in color (and I do refer to the colors), so color printer is better. Or, if you rather not stay connected, you can download it and view on the screen... The artical is educational, but it is not just "old rehash". Some of the material aproaches things from "different angles", though all within the bounds of science and engineering. best regards to all Dan Lavry |
#3
|
|||
|
|||
Sampling Theory
"dan lavry" wrote in message om... Around the end of 2003, on a thread called 384KHz (regarding fast sampling), I promised to post an artical on sampling with some focus regarding the ridiculess 384KHz and 192KHz trend. It took some work, and it is a long paper. If interested, go to www.lavryengineering.com Click on support, and it is the first artical under "white papers". It is called "Sampling Theory". It is a pdf artical and can be downloaded and printed, but the plots are in color (and I do refer to the colors), so color printer is better. Or, if you rather not stay connected, you can download it and view on the screen... Thanks jb |
#4
|
|||
|
|||
Sampling Theory
In rec.audio.pro, (Ralph & Diane Barone) wrote:
Dan, it appears that your ISP is having problems. I get a redirect to fatcow.com saying that the page does not exist. Do you have flash disabled or not installed at all? I have my browser set to "do you want to run such-and-such?" and I usually say no and get to the text I want to read anyway. When I say no on this site, it goes to a download-flash-or-click-here-to-enter page, and if I "click here to enter" and again say no it goes to the fatcow 404 page. But things work fine if i say yes to the "A script is accessing..." prompts. This should be a shortcut to the article without having to go through the flash website: http://www.lavryengineering.com/docu...ing_Theory.pdf I hope it's okay with Larry that I post the link, but then I hear it's always easier to ask for forgiveness than for permission. ----- http://mindspring.com/~benbradley |
#5
|
|||
|
|||
Sampling Theory
dan lavry wrote:
If interested, go to www.lavryengineering.com If interested? Are you kidding? :-) Excellent piece of work, Dan. I would make one suggestion. At the point where you introduce the sinc() it would be good to motivate that by saying that the sinc() is the result of a filter acting on a real impulse that is infinitessimally short where that filter is a perfect lowpass, giving output equal to input for all frequencies below 1/2 the sample rate and zero for all frequencies above it. I'm sure you can find a better way to state it but I think it's important that people understand why you are using it. Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein |
#6
|
|||
|
|||
Sampling Theory
(dan lavry) writes:
It took some work, and it is a long paper. If interested, go to www.lavryengineering.com Click on support, and it is the first artical under "white papers". It is called "Sampling Theory". It is a pdf artical and can be downloaded and printed, but the plots are in color (and I do refer to the colors), so color printer is better. Or, if you rather not stay connected, you can download it and view on the screen... Can you post a direct url to the paper? When I view www.lavryengineering.com, I get a flash detection page, and the link to bypass flash gets a 404 error. Thanks. |
#7
|
|||
|
|||
Sampling Theory
"Paul Rubin" wrote in message ... Can you post a direct url to the paper? When I view www.lavryengineering.com, I get a flash detection page, and the link to bypass flash gets a 404 error. Thanks. I just get a black screen! TonyP. |
#8
|
|||
|
|||
Sampling Theory
Paul Rubin wrote:
(dan lavry) writes: It took some work, and it is a long paper. If interested, go to www.lavryengineering.com Click on support, and it is the first artical under "white papers". It is called "Sampling Theory". It is a pdf artical and can be downloaded and printed, but the plots are in color (and I do refer to the colors), so color printer is better. Or, if you rather not stay connected, you can download it and view on the screen... Can you post a direct url to the paper? When I view www.lavryengineering.com, I get a flash detection page, and the link to bypass flash gets a 404 error. Thanks. http://www.lavryengineering.com/docu...ing_Theory.pdf |
#9
|
|||
|
|||
Sampling Theory
In article ,
Ben Bradley wrote: In rec.audio.pro, (Ralph & Diane Barone) wrote: Dan, it appears that your ISP is having problems. I get a redirect to fatcow.com saying that the page does not exist. Do you have flash disabled or not installed at all? I have my browser set to "do you want to run such-and-such?" and I usually say no and get to the text I want to read anyway. When I say no on this site, it goes to a download-flash-or-click-here-to-enter page, and if I "click here to enter" and again say no it goes to the fatcow 404 page. But things work fine if i say yes to the "A script is accessing..." prompts. This should be a shortcut to the article without having to go through the flash website: http://www.lavryengineering.com/docu...ing_Theory.pdf I hope it's okay with Larry that I post the link, but then I hear it's always easier to ask for forgiveness than for permission. ----- http://mindspring.com/~benbradley That was the problem Ben. Thanks. |
#10
|
|||
|
|||
Sampling Theory
Arny,
http://www.lavryengineering.com/docu...ing_Theory.pdf Thanks very much for doing that. I too have my Internet options set to Prompt, and I refuse to run ActiveX nonsense on my computer. --Ethan |
#11
|
|||
|
|||
Sampling Theory
In rec.audio.pro, (Ralph & Diane Barone) wrote:
In article , Ben Bradley wrote: ... http://www.lavryengineering.com/docu...ing_Theory.pdf I hope it's okay with Larry that I post the link, but then I hear it's always easier to ask for forgiveness than for permission. ----- http://mindspring.com/~benbradley That was the problem Ben. Thanks. You're welcome. Now I only need to apologize to Dan Lavry for calling him Larry. ----- http://mindspring.com/~benbradley |
#12
|
|||
|
|||
Sampling Theory
"Ben Bradley" wrote in message
... In rec.audio.pro, (Ralph & Diane Barone) wrote: In article , Ben Bradley wrote: ... http://www.lavryengineering.com/docu...ing_Theory.pdf I hope it's okay with Larry that I post the link, but then I hear it's always easier to ask for forgiveness than for permission. ----- http://mindspring.com/~benbradley That was the problem Ben. Thanks. You're welcome. Now I only need to apologize to Dan Lavry for calling him Larry. Larry Lavry - that's his evil twin brother who insists that EVERYTHING must be recorded above 192khz! Seriously though, Dan; that was very interesting stuff - thanks for posting it. -- Neil Henderson Progressive Rock http://www.saqqararecords.com |
#13
|
|||
|
|||
Sampling Theory
In article ,
(dan lavry) wrote: Around the end of 2003, on a thread called 384KHz (regarding fast sampling), I promised to post an artical on sampling with some focus regarding the ridiculess 384KHz and 192KHz trend. It took some work, and it is a long paper. If interested, go to www.lavryengineering.com Click on support, and it is the first artical under "white papers". It is called "Sampling Theory". It is a pdf artical and can be downloaded and printed, but the plots are in color (and I do refer to the colors), so color printer is better. Or, if you rather not stay connected, you can download it and view on the screen... The artical is educational, but it is not just "old rehash". Some of the material aproaches things from "different angles", though all within the bounds of science and engineering. best regards to all Dan Lavry A very good article Dan. I especially liked the derivation of the sinc function as the sum of an infinite number of cosine waves. Thank you. |
#14
|
|||
|
|||
Sampling Theory
"dan lavry" wrote in message om... Around the end of 2003, on a thread called 384KHz (regarding fast sampling), I promised to post an artical on sampling with some focus regarding the ridiculess 384KHz and 192KHz trend. It took some work, and it is a long paper. If interested, go to www.lavryengineering.com Click on support, and it is the first artical under "white papers". It is called "Sampling Theory". It is a pdf artical and can be downloaded and printed, but the plots are in color (and I do refer to the colors), so color printer is better. Or, if you rather not stay connected, you can download it and view on the screen... The artical is educational, but it is not just "old rehash". Some of the material aproaches things from "different angles", though all within the bounds of science and engineering. best regards to all Dan Lavry Thanks Dan! |
#15
|
|||
|
|||
Is there science out there that supports human hearing past 20K? ...Sounds
like you believe that to be the case? -Willow "Willow" wrote in message s.com... Dan, Sounds like you're rapping up a topic here....but I would like to know -exactly- your opinion as a designer about the implementation of clock speeds past the nyquist rate. Specifically, since nyquist stands intact, what are the benefits to any rate past this mark? If something does seem to produce a better image of the original (image at lower clock that is at or above nyquist) can you conlude that it is not a function of nyquist but rather an inperfection in device physics that somehow adds/subtracts to image quality? Aliasing etc...? willow "dan lavry" wrote in message om... S O'Neill wrote in message ... Arny Krueger wrote: I'd take issue with this statement, but not with your general issue. The people who dreamed up 192 KHz sampling did know math, engineering, science and technology. The problem is that they weren't driven by a desire for better sound quality at a reasonable price. Well, my answer, in the spirit of the 9/11 commision investigation: I do not recall what I said and have to look back in my files :-) Seriously, regarding Arny's comments: what I meant to say is that those that argue that 192KHz is good do not know math, engineering, science and technology. I thought I was clear about the fact that many of the implementers are simply afraid to rock the boat, be tagged as a "non team player", limit their career advancment potential or simply be fired. I agree that these implementers know math. I guess I was not clear enough. As a guy who's been an engineering VP and had to deal with the Sales and Marketing VP, I can assure you that engineers do not even HAVE brain children (and only have "brains" when someone wants something), and the S&M VP has all the ideas and does all the real work so Mr. Lavry's statement is absolutely correct. An engineer's main purpose is to use math, engineering, science, and technology to extend the schedule beyond next week. I am glad to hear you say it. I tend to agree that most engineers have their nose pointed at a device or a CRT, and they do not look at what happens around them. So the responsibility for the 192KHz crock is on those that run the companies. I am not sugesting that the CEO's, marketing VP's and managment let the engineers decide what to make or market. I am sugesting that the system is broken, when managment decides to go forward without the technical input from the engineers. Or worst yet, they have the input and know it is a crock and proceed with it. I am not going to say which of the 2 it is. I guess it is safer and nicer to think of it is a "broken system". Again, in the spirit of the 9/11 commision investigation: History will (at best) view that redicules temporary trend to 192KHz audio sampling as insufficient material forward to the decision makers :-) Someone should point out to those folks that Moor's law does not apply to the ear. The ear bandwidth does not double every few years :-) And my dog does not hear 192KHz sampling either... BR Dan Lavry |
#16
|
|||
|
|||
Willow wrote:
Is there science out there that supports human hearing past 20K? ...Sounds like you believe that to be the case? You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. That is, though you might not be able to hear a 21 KC note, having the 21 KC components on a 7 KC note may make an audible effect. I have not seen a good well-conducted study show that this is the case, but I haven't seen anything definitive saying that it's not the case either. What Dan is pointing out is that the only thing the higher sample rates buy you is ultrasonic response, and sometimes they buy you ultrasonic response only at the penalty of degraded linearity. So if the ultrasonic stuff is NOT audible, then there is therefore no reason to use the higher rates. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#17
|
|||
|
|||
I can hear bats.
One I recorded this week does not appear to have any sound below 25k. I can hear it on the recording and I can hear it live. When down sampled 5x it sounds pretty pure to me. 24/96 uses disc space so sorry for the 3.7meg download if you want to listen and comment. http://home.comcast.net/~richpeet/bat.wav So do you think I can hear above 21,000? Rich Peet Scott Dorsey wrote in message ... Willow wrote: Is there science out there that supports human hearing past 20K? ....Sounds like you believe that to be the case? You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. That is, though you might not be able to hear a 21 KC note, having the 21 KC components on a 7 KC note may make an audible effect. I have not seen a good well-conducted study show that this is the case, but I haven't seen anything definitive saying that it's not the case either. What Dan is pointing out is that the only thing the higher sample rates buy you is ultrasonic response, and sometimes they buy you ultrasonic response only at the penalty of degraded linearity. So if the ultrasonic stuff is NOT audible, then there is therefore no reason to use the higher rates. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#18
|
|||
|
|||
Rich Peet wrote:
I can hear bats. So can I, and most other people. One I recorded this week does not appear to have any sound below 25k. Say what? I can hear it on the recording and I can hear it live. That has to do with all the content 16 KHz. When down sampled 5x it sounds pretty pure to me. Say what? 24/96 uses disc space so sorry for the 3.7meg download if you want to listen and comment. http://home.comcast.net/~richpeet/bat.wav So do you think I can hear above 21,000? This recording has a ton of content 16 KHz. Have you analysed it with a FFT? If I downsample it to 44/16 it only loses about 2 dB of average level. |
#19
|
|||
|
|||
It must be some sort of harmonic I hear because if I do a FFT filter of all
freqs below 20,000 then I hear nothing. Rich Rich Peet wrote in message news:9xWnc.8347$UQ.462043@attbi_s51... I can hear bats. One I recorded this week does not appear to have any sound below 25k. I can hear it on the recording and I can hear it live. When down sampled 5x it sounds pretty pure to me. 24/96 uses disc space so sorry for the 3.7meg download if you want to listen and comment. http://home.comcast.net/~richpeet/bat.wav So do you think I can hear above 21,000? Rich Peet Scott Dorsey wrote in message ... Willow wrote: Is there science out there that supports human hearing past 20K? ...Sounds like you believe that to be the case? You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. That is, though you might not be able to hear a 21 KC note, having the 21 KC components on a 7 KC note may make an audible effect. I have not seen a good well-conducted study show that this is the case, but I haven't seen anything definitive saying that it's not the case either. What Dan is pointing out is that the only thing the higher sample rates buy you is ultrasonic response, and sometimes they buy you ultrasonic response only at the penalty of degraded linearity. So if the ultrasonic stuff is NOT audible, then there is therefore no reason to use the higher rates. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#21
|
|||
|
|||
Scott Dorsey wrote: Willow wrote: Is there science out there that supports human hearing past 20K? ...Sounds like you believe that to be the case? You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. That is, though you might not be able to hear a 21 KC note, having the 21 KC components on a 7 KC note may make an audible effect. You can also make the argument that the basilar membrane is not a resonant structure and does not use resonant elements that use sinusoids as a decomposition basis but rather decomposes sound neurally in which case all bets are off relative to sinusoidal testing having anything at all to do with what we hear. How that thing works that we hear with is still mostly a mystery despite what has been published. Check out the current thread titled "Basilar membrane mathematics of place theory" in alt.sci.physics.acoustics for a discussion that seems to blow away the myth of understanding that we presume. Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein |
#22
|
|||
|
|||
Chris Hornbeck wrote:
On 10 May 2004 17:31:02 -0400, (Scott Dorsey) wrote: You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. That is, though you might not be able to hear a 21 KC note, having the 21 KC components on a 7 KC note may make an audible effect. The same general argument might apply for recordings intended for further processing, like noise removal, or for transcriptions from antique media. That argument absolutely applies, and in 1990 I was using that argument here to explain why doing click and pop removal in the analogue domain was still more effective than doing it digitally. Finding the click is much easier if you have substantial ultrasonic content because you're basically looking at the second derivative of the signal. It may be easier for processing algorythmi (!) to do their thing with the most complete accounting of the original materials. For durned sure trying to ignore the frequently large ultrasonics of the nasty analog world causes grief in that world. Absolutely. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#23
|
|||
|
|||
Scott,
You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. I don't think a good argument can be made for that at all. Unless Fourier was wrong (hint: he wasn't) all music and sound is comprised of discrete sine waves of varying frequency and amplitude. So for a transient with a fast rise time, the "fast" components are merely portions of sine waves whose frequencies are beyond our hearing range. Some people may be able to hear a bit past 20 KHz, but *nobody* can hear anything even close to 48 KHz which is the basis for using a 96 KHz sample rate. --Ethan |
#24
|
|||
|
|||
thanks for the reply,
Is there science out there that supports human hearing past 20K? ....Sounds like you believe that to be the case? You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. That is, though you might not be able to hear a 21 KC note, having the 21 KC components on a 7 KC note may make an audible effect. I have not seen a good well-conducted study show that this is the case, but I haven't seen anything definitive saying that it's not the case either. Yeah, i have hear that theory before. It is actually easy to test for with a few recorded transients @ 48 and up-sample to 96. Then record same transiets at 96, Blind test those results....Anyone game? What Dan is pointing out is that the only thing the higher sample rates buy you is ultrasonic response, and sometimes they buy you ultrasonic response only at the penalty of degraded linearity. So if the ultrasonic stuff is NOT audible, then there is therefore no reason to use the higher rates. Actually there are more trade off than those mentioned. Doesn't Dan build a box that sample up to 96 and why not 192 or 384 etc....? I have not read the white paper yet, I would assume he answers these q's there. Regards Willow |
#25
|
|||
|
|||
Willow wrote:
Is there science out there that supports human hearing past 20K? ...Sounds like you believe that to be the case? You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. That is, though you might not be able to hear a 21 KC note, having the 21 KC components on a 7 KC note may make an audible effect. I have not seen a good well-conducted study show that this is the case, but I haven't seen anything definitive saying that it's not the case either. Yeah, i have hear that theory before. It is actually easy to test for with a few recorded transients @ 48 and up-sample to 96. Then record same transiets at 96, Blind test those results....Anyone game? How do you know you aren't testing some difference between the way the A/D converter works at the two different rates? Or some artifact of the SRC? The problem is that there are so many of these compounding variables involved. What Dan is pointing out is that the only thing the higher sample rates buy you is ultrasonic response, and sometimes they buy you ultrasonic response only at the penalty of degraded linearity. So if the ultrasonic stuff is NOT audible, then there is therefore no reason to use the higher rates. Actually there are more trade off than those mentioned. Doesn't Dan build a box that sample up to 96 and why not 192 or 384 etc....? I have not read the white paper yet, I would assume he answers these q's there. Yes. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |
#26
|
|||
|
|||
Willow wrote:
thanks for the reply, Is there science out there that supports human hearing past 20K? ...Sounds like you believe that to be the case? You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. That is, though you might not be able to hear a 21 KC note, having the 21 KC components on a 7 KC note may make an audible effect. Change the waveform, then for sure you change the spectral content. Above 500-1000 KHz, the ear is very spectral-centric. I have not seen a good well-conducted study show that this is the case, but I haven't seen anything definitive saying that it's not the case either. The proof is in the filtering: http://www.pcabx.com/technical/low_pass/index.htm Yeah, i have hear that theory before. It is actually easy to test for with a few recorded transients @ 48 and up-sample to 96. Then record same transients at 96, Blind test those results....Anyone game? Been there done that, except at a variety of sample rates: http://www.pcabx.com/technical/sample_rates/index.htm What Dan is pointing out is that the only thing the higher sample rates buy you is ultrasonic response, and sometimes they buy you ultrasonic response only at the penalty of degraded linearity. So if the ultrasonic stuff is NOT audible, then there is therefore no reason to use the higher rates. With ultrasonic response comes lots of ultrasonic noises. For example the recording of a Bat I just downloaded lately had an interesting agglomeration of noise above 30 KHz that didn't seem to vary with the bat. Actually there are more trade off than those mentioned. Doesn't Dan build a box that sample up to 96 and why not 192 or 384 etc....? All you need is Audition, it resamples up to 100 MHz. Seems like high that would be high enough... LynxTWO series audio interfaces among others, perform well at up 192 KHz sampling. I have not read the white paper yet, I would assume he answers these q's there. Listen for yourself! |
#27
|
|||
|
|||
Scott Dorsey wrote:
Willow wrote: Is there science out there that supports human hearing past 20K? ...Sounds like you believe that to be the case? You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. That is, though you might not be able to hear a 21 KC note, having the 21 KC components on a 7 KC note may make an audible effect. I have not seen a good well-conducted study show that this is the case, but I haven't seen anything definitive saying that it's not the case either. Yeah, i have hear that theory before. It is actually easy to test for with a few recorded transients @ 48 and up-sample to 96. Then record same transients at 96, Blind test those results....Anyone game? How do you know you aren't testing some difference between the way the A/D converter works at the two different rates? Point taken. The route I took at www.pcabx.com was to record at 96 KHz, and then downsample to the test sample rate and then upsample back to 96 KHz. Or some artifact of the SRC? Run technical tests on the SRC. The problem is that there are so many of these compounding variables involved. Got more? If the listening tests come out "no differences" then the alleged confounding influences must not be all that significant. Or, somehow they managed to exactly cancel out the effect of the difference in sample rates. Seems pretty darn unlikely! |
#28
|
|||
|
|||
Ethan Winer wrote: Scott, You can make a good argument that, even though pure tones over 20KC are not audible, that the added bandwidth makes transients more accurate. I don't think a good argument can be made for that at all. Unless Fourier was wrong (hint: he wasn't) all music and sound is comprised of discrete sine waves of varying frequency and amplitude. A time domain signal is what it is, no more and no less. If you have a very particular kind of hardware that can perform correlations of it with complex exponentials then you can get as an output a domain transformation in terms of complex frequency that was first noticed by Fourier and can be useful for computation. There is no evidence whatsoever that the ear/brain performs that kind of transformation and it is so unlikely compared to more useful ones in the evolutionary sense that it should be entirely discarded as a possibility. OTOH, there are decomposition basis classes that seem more likely which would have relatively small responses to sinusoidal stimuli, if any in certain ranges. The fact that the ear/brain fails to respond to sinusoidal stimuli in a frequency range says almost nothing about how it will respond to acoustic stimuli which, if analysed in the Fourier sense, would show harmonic content there. Discussions of what we can or can't hear that are argued on the basis of Hz are in fact baseless in a very literal sense. That's not to say that experimental work such as Arny has done with determining what lowpass corner renders real acoustic data perceptually different isn't extremely valuable because, of course, physics says that there will be some upper corner above which the physical mechanism cannot respond in any way but let's just forget about how it might respond below that limit and how that relates to sinusoidal hearing tests. Bob -- "Things should be described as simply as possible, but no simpler." A. Einstein |
#29
|
|||
|
|||
The most important fact we have is the most obvious one; sound is pressure
variations, and our hearing mechanism, starting from the pressure variations arriving at the eardrum, is not brickwall-filtered. We feel 0.5 Hz, and our brain reacts to 70kHz. Even if it's subconcious and we claim "we can't hear it", the energy doesn't mystically disappear. These kind of frequencies affect the naturalness of the sound, but their importance is another matter... Personally, I'm happy working without the ultrasonic range. The whole ultrasonic range brings too many problems compared to its benefits. That is, in audio reproduction. |
#30
|
|||
|
|||
Bob and Tommi,
I think you're both making this more complicated than necessary. A fast rise time is just that, and if there are 70 KHz components we don't hear them. I can't imagine how we could sense them either since they'd have to get through our ears to reach our brains. The ear and brain don't "transform" anything. They simply respond to frequencies that aren't so high the mechanical parts can't vibrate. As you acknowledge, low-pass filter tests are absolutely valid, and my own tests many years ago convinced me that this is all one huge non-issue. If disk space and CPU power were free, then sure, record at a 20 MHz sample rate for all I care. But for now, anyway, these resources are not free, and that's the real issue in practical terms. --Ethan |
#31
|
|||
|
|||
Tommi wrote:
The most important fact we have is the most obvious one; sound is pressure variations, and our hearing mechanism, starting from the pressure variations arriving at the eardrum, is not brickwall-filtered. We feel 0.5 Hz, and our brain reacts to 70kHz. The half hertz I get, but the 70 KHz escapes me. Please explain. Even if it's subconscious and we claim "we can't hear it", the energy doesn't mystically disappear. There's nothing mystical at all about how it disappears. But, that doesn't mean that anybody hears it. These kind of frequencies affect the naturalness of the sound, but their importance is another matter... That would be an unproven assertion, even an assertion that escapes proof when diligently pursued. Personally, I'm happy working without the ultrasonic range. So is anybody who is interested in that which is heard, as opposed to that which is supposed. The whole ultrasonic range brings too many problems compared to its benefits. Open the windows wider, and more dirt comes in. That is, in audio reproduction. |
#32
|
|||
|
|||
The ear is not perfectly linear. Ultrasonic components _could_ intermodulate to
produce audible components. (There's a speaker system based on this principle.) Whether or not this actually occurs with ordinary musical sounds is something I don't know. |
#33
|
|||
|
|||
I don't mean to steal a thread here but as I was testing a higher sample
rate to begin with: If you are talking about the bat from this thread I noticed the noise too. The mic was pointed toward an expressway about 1/2 mile out so I suspect that as the source. Would you agree that the noise appears to be environmental? The reason that is important to me is that then I can use that noise to determine if my mic is responding fairly well from 1 to 40,000 cycles with a few lower frequency bands of trouble. Testing the system was the original goal here. The loud noise is a stone I tossed up to attract the bat (works every time) which then bounced off the roof. recording chain was: Senn mkh-110 (unbalanced),homemade mic power supply, rme quadmic pre, rme multiface, Audition 1.0, toshiba laptop with xp. Rich Peet Arny Krueger wrote in message ... .... With ultrasonic response comes lots of ultrasonic noises. For example the recording of a Bat I just downloaded lately had an interesting agglomeration of noise above 30 KHz that didn't seem to vary with the bat. Actually there are more trade off than those mentioned. Doesn't Dan build a box that sample up to 96 and why not 192 or 384 etc....? All you need is Audition, it resamples up to 100 MHz. Seems like high that would be high enough... LynxTWO series audio interfaces among others, perform well at up 192 KHz sampling. I have not read the white paper yet, I would assume he answers these q's there. Listen for yourself! |
#34
|
|||
|
|||
William Sommerwerck wrote:
The ear is not perfectly linear. Ultrasonic components _could_ intermodulate to produce audible components. (There's a speaker system based on this principle.) ...snip.. I think I've missed something. Are you saying the ultrasonic speaker system uses the ear's non linearity, or are these just two (somewhat) isolated facts? I believe the speaker system you are talking about produces sound emanating from remote locations in space through nonlinear interactions of ultrasonic beams with air. I don't believe ear linearity has anything to do with that system. Buy the way, the physics that allow that "speaker system" to work also relate to the acoustic refrigerator... :-) [ Vortex cooling is another story. ] Later... Ron Capik -- |
#35
|
|||
|
|||
William Sommerwerck wrote:
The ear is not perfectly linear. Right, nonlinearity sets in pretty strongly above 75-85 dB. the effect is strong enough that audible differences in DBTs can "go away" if the volume is too high. |
#36
|
|||
|
|||
Rich Peet wrote:
I don't mean to steal a thread here but as I was testing a higher sample rate to begin with: If you are talking about the bat from this thread I noticed the noise too. The mic was pointed toward an expressway about 1/2 mile out so I suspect that as the source. Would you agree that the noise appears to be environmental? Let me put it this way, the shape of the lump is all wrong for the usual switchmode power supply noise, etc. The reason that is important to me is that then I can use that noise to determine if my mic is responding fairly well from 1 to 40,000 cycles with a few lower frequency bands of trouble. Testing the system was the original goal here. The loud noise is a stone I tossed up to attract the bat (works every time) which then bounced off the roof. OK, that explains some things. recording chain was: Senn mkh-110 (unbalanced),homemade mic power supply, rme quadmic pre, rme multiface, Audition 1.0, toshiba laptop with xp. You might want to look at my related stuff posted at: http://64.41.69.21/technical/sample_rates/index.htm Lots of spectrograms, spectral analyses, and downloadable samples @24/96. Equipment list is there, too. |
#37
|
|||
|
|||
"Arny Krueger" wrote in message ... Tommi wrote: The most important fact we have is the most obvious one; sound is pressure variations, and our hearing mechanism, starting from the pressure variations arriving at the eardrum, is not brickwall-filtered. We feel 0.5 Hz, and our brain reacts to 70kHz. The half hertz I get, but the 70 KHz escapes me. Please explain. I bet most of you guys have read already read this, but: http://www.sowter.co.uk/pdf/ultrasonichearing.pdf These kind of frequencies affect the naturalness of the sound, but their importance is another matter... That would be an unproven assertion, even an assertion that escapes proof when diligently pursued. Most cymbals have over 40% of their energy in the 20kHz+ range. |
#38
|
|||
|
|||
Tommi wrote:
"Arny Krueger" wrote in message ... Tommi wrote: The most important fact we have is the most obvious one; sound is pressure variations, and our hearing mechanism, starting from the pressure variations arriving at the eardrum, is not brickwall-filtered. We feel 0.5 Hz, and our brain reacts to 70kHz. The half hertz I get, but the 70 KHz escapes me. Please explain. I bet most of you guys have read already read this, but: http://www.sowter.co.uk/pdf/ultrasonichearing.pdf Yes, been there, deconstructed it. These kind of frequencies affect the naturalness of the sound, but their importance is another matter... That would be an unproven assertion, even an assertion that escapes proof when diligently pursued. Most cymbals have over 40% of their energy in the 20kHz+ range. I'm not sure of that. I've seen a lot of 24/96 cymbals recordings with energy peaking in the low teens. The word cymbals covers a lot of territory. Take a look at http://www.pcabx.com/technical/sample_rates/index.htm The keys recording has a similar amount of energy between 22 and 48 KHz. But when you do the DBT listening tests... |
#39
|
|||
|
|||
In article ,
"Tommi" wrote: "Arny Krueger" wrote in message ... Tommi wrote: The most important fact we have is the most obvious one; sound is pressure variations, and our hearing mechanism, starting from the pressure variations arriving at the eardrum, is not brickwall-filtered. We feel 0.5 Hz, and our brain reacts to 70kHz. The half hertz I get, but the 70 KHz escapes me. Please explain. I bet most of you guys have read already read this, but: http://www.sowter.co.uk/pdf/ultrasonichearing.pdf Yes, but alpha wave production is a rather crude way of establishing what is going on here. Just closing your eyes can increase alpha production by 50% or more. It is evident that something is being picked up by the brain but there's no physiological explanation for the effect observed. If I saw auditory nerve firing in response to ultrasonic input I'd consider it hearing. This is probably, as the authors mention, a modulation effect, but it may well be important in perception. These kind of frequencies affect the naturalness of the sound, but their importance is another matter... That would be an unproven assertion, even an assertion that escapes proof when diligently pursued. Most cymbals have over 40% of their energy in the 20kHz+ range. Virtually every instrument we've tried has some energy above 20 kHz. -Jay -- x------- Jay Kadis ------- x---- Jay's Attic Studio ------x x Lecturer, Audio Engineer x Dexter Records x x CCRMA, Stanford University x http://www.offbeats.com/ x x-------- http://ccrma-www.stanford.edu/~jay/ ----------x |
#40
|
|||
|
|||
William,
The ear is not perfectly linear. I'm fully aware of this, and to me it's a completely different issue. I sometimes play percussion in a local symphony, and that includes orchestra bells. When playing two notes very loudly I can easily hear the IM products that are generated inside my ears. How close musically the notes are to each other determines which low frequency beat note I hear. I agree that some of the beating is between ultrasonic components. But this does not mean an audio system needs to capture ultrasonics! Moreover, this IM effect is not audible out in the audience, and even I hear it only when playing very loudly. To take this to the logical conclusion, the "ultrasonics matter" camp would have us capture everything up to 48 or 96 KHz or even higher, and they'd have us buy loudspeakers that can reproduce that high, merely so we can play it really loud while standing next to the speakers so our ears can generate IM distortion. No thanks. --Ethan |
Reply |
|
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
For real-world ADCs, is sampling done instantaneously, or averaged? | High End Audio | |||
Digital high frequency distortion | High End Audio | |||
More digital theory: Parallel AD? | Pro Audio | |||
Why 24/96 sampling isn't necessarily better-sounding than 24/44 sampling | Pro Audio | |||
FS: Roland S-50 Sampling KB w/Extras | Pro Audio |