gubu

Actually it originates at the source (instrument/microphone). But that proves your point how? I'm gonna repeat it like a mantra from now on:- Math does not give you better time domain resolution. Only a higher sample rate can do that.

The voltage comes from the coffee grinder. No buddy, the voltage comes from the input. At least when I'm being facetious, i stay on topic. Math doesn't give you better time domain resolution, only a higher sample rate can do that. The weak link in the chain is always something simple and obvious.

All waves are sine waves. A square wave is just a few sinewaves added together and so on. So there's no voltage being sampled at all? It's all just math is it? Also, math doesn't give you higher time domain resolution, only a higher sample rate can do that

Recording analog with tape running high speed captured more details than a tape running consumer speeds. This allowed for a high fidelity working enviornment that cause minimum losses during mix and mastering processing. This is not much different working in digital. The effects of processing digital samples is different but the goal ultimate goal is the same.

"sampling at 44.1k produces square waves. And that's a fact!" Sincerely that is funniest BS I've hear in long time. And I've heard some doosies. I honestly have no idea how you come to that conclusion, all I can imagine is somehow you think this is like connect the dots, with a straight edge. The math of sound is based on complex addition/ subtraction of sinewaves not straight lines. Because sinewaves are symmetrical, all you is need two points anywhere within its period and you can determine everything you need to redraw it. Put this into your calculator then:- How does a convertor running at 44.1kHz know whether a 20kHz waveform is sinusoidal, square, sawtooth or Mickey Mouse doing a tap dance when it's only sampling the voltage 2 times per cycle? I'm sure the complex math gives you back some shape of a waveform, just not necessarily the waveform you put in. Until someone who actually develops and designs convertors from scratch comes on here and tells me different, I am not accepting that 44.1kHz or 48kHz samples frequencies at the upper limit of human hearing as accurately as 96kHz or 192kHz. I've done A/B recording tests and there is an audible difference, so you guys can tell me what you think I heard and tell me that I'm full of {censored} from now til the end of time, the map is not the waveform, no matter how convoluted the math is in sampling it and reproducing it. More accuracy in the time domain, for that's what a higher sample rate gives you, produces a more accurate reproduction of the waveform. Good luck to ye

It isn't analogous to 1" and 2" tape at all. It isn't even similar. As to why it was developed, well, you'll have to ask the guys in the marketing department. I've done numerous A/B checks recording at rates from 44.1kHz to 96kHz, and if you have a decent converter it doesn't make a single bit of difference. Not even a small one. edit: You keep saying that time is a factor. Of course it's a factor. Why do you think we keep talking about frequency? I've been talking about nothing but frequency as well. I'm sure there were plenty of folks who said that 2" tape was a marketing excercise when it first came out too. Higher sample rate = more accurate sampling of frequencies approaching the upper limit of human hearing. Read the Master Handbook of Acoustics. I'm not some internet nark, I get my information in the real world.

That was quite nasty, sorry guys. Do google that book tho, I know I'm not talking garbage and I can guarantee you that if you do an AB recording session using a 48kHz and a 96kHz system, the 96kHz system recording will sound better, even bounced down to disk @16 bit 44.1Khz. The difference may be small, but then why bother developing 192kHz systems if it makes no difference? It is analogous to the difference between 1" and 2" tape.

Wow. Bamboozling ourselves with mathematics? Now who's insulting someone's intelligence? You're using just enough science to back up what you're saying without applying enough of it to realize that you're drawing an erroneous conclusion. Believe what you want to believe. But converters and signal manipulation are nothing BUT mathematics. Until you accept that, you're going to have a hard time getting anywhere. Yes of course they are nothing but mathematics but even a child can see that 2 samples per waveform = a square wave. That's not erroneous. If you can't see that time is a factor in this math, I can't help you. I'm basing this on reading the Nyquist theorem in school and further study in 'The Master Handbook of Acoustics.' Google it. It sounds like you guys could all do with buying a copy.

Guys, you can bamboozle yourselves and everyone else with mathematics til the cows come home. The simple fact is that any waveform sampled at a rate only twice it's it's frequency is going to produce a squared off version of the original waveform. The output convertor is feeding a voltage value to the op amp 44,100, 48,000 or 96,000 times a second and 2 samples per waveform = a square wave. Convertors may have something built in that makes the waveform more sinusoidal but you're still not getting out what you put in. I've no doubt it's difficult to hear the difference between a SACD and a regular CD in AB tests but that has absolutely nothing to do with recording. If you record and mix in 96kHz you will get a far more accurate sampling of the frequencies approaching the upper limit of human hearing than in 48kHz or 44.1kHz. And that's before you start adding processing, which as pointed out above, makes things even more grainy. That graininess is a squaring off of the waveform. It's not going to make much of a difference for lower frequencies as there are enough samples per waveform for it not to be noticable but heading towards 20kHZ, it does make a difference and this is where all the spatial and depth information is in a signal. It has absolutely nothing to do with frequencies outside the range of human hearing phasing with the ones we can hear or any other nonsense like that, it's a simple case of getting out what you put in.

BTW, I did Nyquist in school too and all I learned from it is that if you're gonna make an AD convertor, you'd better make sure to stick a LPF in there somewhere to kill all freqs above half the sample rate unless you want everything to sound like internet streaming audio circa 1996. How does a 20kHz waveform not become square if you're sampling at 40kHz? Do the little elves make it back into the same shape it was when it went in? Sorry for the deliberate facetiousness but come on guys....

mozart, grandma, and math have nothing in common. That's what you think!!! Hey guys, thanks for the chat, I've gotta get some zzzzz time take care ad

Nope. There's a whole field of engineering and mathematics behind this, and I've only taken a couple of classes (and college was years ago so I've forgotten quite a bit of it), but once you've reached that 2-samples-per-cycle number, you get an accurate waveform. If you're interested, read up a bit on the Nyquist sampling theorem. Or Nyquist-Shannon, I believe it's called now. Wikipedia's Entry What it basically boils down to is, an analog signal (such as a soundwave, or more accurately, the electrical signal caused by that soundwave hitting a microphone diaphragm) can be represented as a sum of sinusoidal components. Any sinusoidal component can be accurately sampled as long as the sample rate is above twice the frequency of the sinusoid. Thus, any signal can be accurately reproduced as long as the sampling rate is above twice the highest frequency of the signal. Whew. That's very long-winded, isn't it? There's a fair bit of math behind it, and it took me quite a while to see it too, because it seems counterintuitive - it seems like more samples OUGHT to make for a more accurate waveform. But it doesn't. It just makes for a waveform that's just as accurate as the one we already had. So what you're telling me is that an 88Hz bottom E on a guitar can be accurately sampled at 176hZ? I wouldn't think so now to be honest, it's all going to go a bit square isn't it, regardless of the harmonics. Look, we'll have to agree to differ, if I ever win big on the lottery, I'm going 96KhZ, as I said above, it's like the difference between 1" and 2" tape, it's just a better format if you can afford it

Sure its better, but your ears can't hear it, your microphones can't capture it, and your speakers can't play it. The mics capture the waveform up to their threshold, that's fine but what you've said there is like saying that there is no difference between Mozart and my grandmother when you play them thru a transistor radio. I'm sorry sir, but there is...

You might rather, but your software doesn't care. It's going to reproduce the same waveform from the 2 samples as it would from the four. It can't possibly do so. And ok maybe it might, I'm not a boffin, but how about 10 samples per 9kHz waveform than 5, that's gotta be a better representation of what went in, no?

OK, I'm afraid this just isn't true. You'll capture the exact same waveform in this scenario at 44.1kHz, 96kHz, or even at 20kHz. You need a sampling frequency equal to twice the highest frequency you want to capture. That's it. Anything above that will not improve the accuracy of the waveform in any shape, form, or fashion. Yes it will, time is a factor. I'd rather have 4 samples per 20kHz waveform than 2

But that's just what I'm saying - no offense intended, but 96kHz will only give you a "more accurate map of the waveform" if you're trying to capture frequencies above 24kHz, and you shouldn't have any of that bouncing around anyway. At anything in the actual audible range you'll get the same waveform from 44.1kHz as you'll get at 192kHz. EXACTLY the same. I apologize if the first post came across as insulting - it wasn't intended to be. Sometimes text doesn't do a very good job of conveying a tone of voice. Furthermore, if you did start capturing frequencies in the 40kHz range ... what then? Microphones generally list their frequency ranges as being in the 20-20kHz band, and even if you have monitors that will reproduce sounds above 20kHz your audience certainly will not unless you distribute exclusively to rabid audiophiles. The point of higher sampling rates relates purely to the hardware design of the converters, as I said before - it simply makes it easier to design a low-pass filter (that filters out any noise at inaudible high frequencies) to put in front of the converter. That's it. On a good, properly-designed converter, there isn't much point in using anything above 44.1kHz. I'm not talking about frequencies that we can't hear. If you set a LPF to 10kHz and record in 96kHz, you are still getting a more accurate encoding of the waveform over time and it is audible. For making CD's, it probably doesn't make any sort of qualitative difference, but if I had the money, I know how I'd be recording.

96kHz doesn't give you a better recording, unless you're using converters with a fairly poorly-designed low-pass filter on the high end. It just doesn't. Regardless of what anyone thought they heard on any A/B tests, or any other anecdotal evidence to the contrary. I didn't think I heard anything, it sounded better in 96kHz on playback thru both systems. I'm not telling anyone to go off and waste money on a system that can handle 96kHz when 48 or 44.1kHz works just fine for anything that's going onto CD. Filters aside, 96kHz is still a more accurate map of the waveform whichever way you look at it. Don't insult my intelligence like that again please friend.

You've heard wrong. Sampling rate governs the reproduction of frequency, and 44.1kHz will reproduce the full range of human hearing. There is some controversy about the interaction of inaudible high frequencies with the audible ones, but the overwhelming consensus, backed up by the failure of any study to demonstrate a contrary position, is that 44.1kHz or 48kHz will do anything that 88kHz and beyond will do, at half the memory and disc usage. Not true. Myself and another guy ran the same mic into 2 identical Protools systems as a test a bout 3 years ago. The systems were both Digidesign 002's into the same version of PTLE on Macs. We got a lend of a 1 in, 2 out xlr splitter from another friend and strummed a guitar for a while into a Se Gemini mic. We ran the test repeatedly running 48kHZ and 96kHz on both machines and thru both sides of the splitter box just to be sure. The recordings in 96kHz returned a markedly more spacious, defined and 'deeper' recording of the source on playback thru a pair of Genelec monitors. I still use 48kHz on everything for ease of memory and disk usage and it sounds great but what you are saying there about higher sample rates is total falsehood, 96kHZ gives you a better recording if you can afford the hardware.

I've heard drums are especially suited for 96kHz rather than 48kHz. Would this pertain to analog drum machines as well? How about voices as well, do they benefit much from the increased sample rate? It's the same as the difference between one inch and two inch tape in the old days. Bigger is better, you have a more accurate map of the waveform pure and simple. Bigger sample rate also means more cost with regard to convertors, storage, processor power etc etc just like it did in the old days with 2" machines being more expensive to buy and maintain than thinner formats.