So to me this has a somewhat obvious answer, but I've never actually heard it discussed so I'm not certain, and I'm not sure many people have thought about it too much.
When working with a signal chain of say, just a microphone, pre-amp, and recording device, the maximum frequency response of the end-result file is going to be the frequency response of the lowest response in the chain, correct?
Let's say I have a Sound Devices 722 two channel recorder - which has a FR of 10hz to 40khz, and a Senheisser MKH800, which goes up to 50khz. Even though the sample rate of the 722 goes up to 192khz, the file recorded on the SD722 is only going to reproduce up to 40khz, correct?
The flipside to this is - if I have a Sanken CO-100K, with a response of 20hz - 100Khz, and a Korg MR-1000, which also has a frequency response for 20-hz to 100Khz, I should be getting a much better file than the 722 using a sanken CO-100k, right?
The big problem is that I can't think of any way to test a lot of this stuff - there's no frequency analyzers that I know of that go up to 100k, and I don't want to buy an oscilloscope.
So, what's ya gonna listen too above 12khz?
Well, I still have most of my hearing, so I'll be listening to things between 20hz and 18-20ish khz :P
Mainly I want to get up near 100k for sound effects recording - take a 192khz file with a true 96k-100k response frequency, and rewrite the header information so that the file thinks it's a 96khz or 48khz sample rate file. It'll playback at half or quarter speed, with all the frequency shifted down into the audible range. IE, at 96k, things recorded at 40khz will be heard around the 20khz range. Supposedly. I've only done one brief test, with the Korg Mr. 1000 and the sennheiser MKH800, but it seemed to work pretty well.
When was the last time you had your hearing checked by a professional audiologist?
Just because the converters go that high doesn't mean the rest of the electriconic do.
OK, first off, if you don't want to be helpful, leave thee thread. I don't need you here getting on my nerves, and your responses seem to be formed specifically to do that.
One, I never said it was just the A/D converters that have those frequency responses. I'm guessing that if a device has a rated frequency response, that's what the OVERALL device can reproduce, not just some piece of it's electronics.
Secondly, what the hell does it matter how well I can hear? I want to record things that high, even if I can't hear them back at realtime, I can still work with it.
Give me a real answer, or contribute to the question in an actual meaningful way, or go away.
There most certainly are hardware measuring tools that will measure up to 100k and well beyond. Whether you can afford them is a different matter.
Even if you could measure it right now, what would it mean? It changes nothing and you can't change anything about it even if you know the measurement result or not.
What is there to work with if your not hearing it? Being able to measure or see a plot of vibrating coliding atoms and molocules is not audio or music? What would be the point in wasting time with that?
...he wants to play it at reduced speed, bringing say, 80KHz to 10KHz...
Yes, what Codemonkey said. Eventually I might try and get a hold of some Murata super-tweeters, but for right now I just want it for sound effects generating, and extremely reduced playback speeds.
Rewrite the header information of a 192k file, so that it thinks it's a 96k file, and you're effectively playing back at half-speed, without losing any actual data. Keychains sound like massive chains for a drawbridge, etc.
You have to be a little careful when considering the concept of frequency response in relation to human audio perception. Note that I didn't say "hearing", as our ears tell us more than we think we hear.
Frequency response is essentially a steady-state measurement, which in practical terms means that a single tone at any test frequency has to be present long enough for a meaningful measurement to be made. If all we cared about in audio recording and reproduction is the frequency response, we would have a very dull experience.
As well as the steady-state frequency response, the human auditory system has a very well evolved response to transients. The transient response of systems is usually specified in the time domain and not the frequency domain, and so it's easy to miss the fact that the equivalent frequencies extend well above the range characterised by the steady-state frequency response. Transient response also ages differently from frequency response, so even older people who on normal frequency-based hearing tests would be classed as deaf above (say) 12KHz can tell the difference on a transient test between equipment having a poor and an extended high-frequency response.
What this comes down to is that higher sampling rates do have their place in audio recording, and it could well be the reason that people can tell the difference between brick-wall band-limited digital audio and an analog recording, where both have the same nominal -3dB high-frequency point.
In reply to a point in your original post where you ask whether the overall frequency response of a signal chain is limited by the narrowest component, the answer is yes, but..
In system terms, the frequency responses are multiplied together point-by-point, so indeed the narrowest component may well have a dominant effect, but all the others will contribute too. Two cascaded responses of -3dB at 20KHz do give -6dB at 20KHz, but the roll-off before 20KHz is affected as well.
However, the transient response of the chain will be much more affected by the type of equipment in that chain. Cascading two pure analog components each having a nominal upper 3dB point at 20KHz will sound better for music than cascading two digital components (via their analog ports), not just because of the extra A-D and D-A processes, but because of the brick-wall filters needed for Nyquist anti-aliasing purposes, and their detrimental effect on the transient response. Higher sampling rates push the Nyquist frequency correspondingly higher and so allow better reproduction of transients.
Your title question does not have "an obvious answer"!
I would like to inject that just because an analog piece of equipment is rated at 10Hz-20kHz it does not mean that no signals (frequencies) are passed above or below the frequency rating. It only means that it rolls off below a certain threshold and at that point it becomes less effective with the frequencies above or below its rating.
I think this was probably stated above in much more technical terms but I thought I would say it anyway because it is an interesting topic.
Bottom line is the bottom dollar.
You can record anything you want to record at any frequencies you want to record at. Provided you have the bucks to accomplish your 100kHz + recordings.
Sure, many pieces of equipment make it up to and/or beyond 100kHz. I know that I can't hear beyond 15kHz at 53 years of age but I still perceive frequencies above that consistently. Go figure? It's our hearing apparatus and brain interface. So like Boswell said, we can hear things that cannot be explained fully.
Your biggest dilemma is all of the RF junk you'll get with frequency responses beyond 20kHz. Don't forget that 100kHz actually equates to a radiofrequency. You're just not making sense just because you think something should be a certain way. That's ridiculous. You don't need to record at 192kHz and downsample to 96kHz to record something, just to pitch it down. Software offers a myriad of ways of doing this without having to record at 192kHz. So what the heck do want to record bat farts while they hick-up? What an accomplishment!
Why don't you just learn your tools and record what you want? You have to have some kind of education to do what you want to do which you have none of. Might I suggest going to Bat Fart University?
Ms. Remy Ann David
President of Bat Fart University's Bat hick-up audiblility division.
Sounds like mushoo's wanting to ride the electrons...
Let's saddle up... Hang on... here we go...
While you would think that your logic was "sound" that the narrowest frequency response'd device is the maximum frequency response captured... I must respectfully disagree with Boswell that it's not.
The frequency response captured is the maximum frequency response of the recording device.... period. Whether any signal is present from the rest of the chain indeed dependent upon the narrowest device in the chain... and on that point I concur with Boswell. In other words, if your recording device is 10Hz to 100kHz, then the response is that wide... no wider, no narrower.
If it's 20Hz to 20kHz, then it makes no difference if anything (or everything) in the chain is triple the bandwidth of your device. The device will only capture it's maximum bandwidth.
Thanks, MadMax, that's about what I'd figured. The MR-1000 has a nice little toggle-able filter for DSD formats, specifically to cut out or avoid any radio frequency noise in the top end - I _have_ been able to test that a little bit, as my recordings have a hiss at around 10-15k when played back at quarter-speed and the filter off, and it's gone with the filter on.
So yes, it does record things far beyond the range of whatever the mic I might be using is (I wonder too, how well they've got it shielded from itself - IE, how much of it's own internal electronics is it picking up?)
And Ms. Remy, you are one of the most annoying writers I've ever read. Why do you assume I have no education? Who says I want to record bat farts? I'm going to a machine shop to record industrial water-jet cutters slicing through various objects, and I want to slow down playback to the point that I can hear what's going on well beyond my normal hearing range.
And it's not downsampling, it's tricking the WAV file to playback at half speed WITHOUT running it through a pitching algorithm of ANY sort - the less filtering I have to do to something, the better.
can I give you half an expect whole participation?
Had you said that in the original outing....
edit: and this is at least a small portion of why things become what they become in the forum(s) as being discussed over here:
it doesn't help you hear
I have a LinearX ISA-Vesa card that can measure up to 100k
not good in machines above 25mhz
I am still running a 386 with co-proccessor to keep this card running
the 286 finally failed
and to cap it off the processor and the co-processor is stamped AMD/Intel ... !?
I wouldn't be without a quality ANALOG oscilloscope for looking at this stuff and double checking the digital gear
Of course people have thought about this
been there done that
testing the ellectronics is not so hard
testing speakers or transducers to 50k and above is do-able as you can get test mics with calibration files relatively cheaply
testing mics to beyond 25k is not so easy as you need a reliable predicatble transducer to create the test tones
you need an environment for doing the test.
these tests need to be repeatable and probably need a couple of methods to get a good result
not just steady state but FFT and Max length stuff
then compare and average all results
can be fun and is a good learning exercise
I'm a day late and a dollar short. But I gotta get some weight in here.
My Behringer mic picks up sound outside of it's frequency response. I recorded my voice, put a low pass filter on it, set to about 10hz below it's stated cutoff, amplified the result and played it through a sub. There it was, massaging my head, dirty low bass.
Oh, and Mooshoo, you are way more annoying than Ms. David.