Higher sample rates are often used by audio equipment manufacturers to make themselves look better - because of course, the higher the number, the better it is, right? Right?!
But what do samples rates actually mean for producers and engineers? Which sample rate should we be using, and is it better to use higher sample rates when recording?
Let's start at the beginning - converting our electrical (analogue) signal into a digital one that our computers can read.
simply put, an ADC reads the volume level of a wave at a series of points. Each blue X on the wave below represents a sample.
Now our first thought here is "if I take more samples, I get better definition", and that's kind of true, but what really changes when you change the sample rate is how many frequencies you can effectively capture.
Think about it like this - if I record a 20kHz sine wave at 20kHz sample rate, every sample will be taken at the same point in in the sine wave’s cycle, so, depending on the timings my audio signal will be either a flat 0, full scale DC, or anything in between.
If I sample the same wave at 40kHz, I have 2 samples per wave cycle. This could work! Most of the time I’d capture the signal between the positive and the negative parts of the cycle, so I should record that signal. However, if by chance the sample point happens to fall when the sine wave crosses zero - bugger! We’ve got two zero samples and our sine wave disappears again!
This is how in general we came to that 44.1kHz value. Because the maximum frequency most humans hear is 20kHz, if we make sure we take enough samples to actually reproduce that wave, and make sure that we’re never in perfect sync with that wave, then all’s good!
This brings us to our next issue - if we’re just sampling away and we send a sine wave at, say, 80kHz - we’re not going to get an accurate reading of an 80kHz wave. But we are going to see something recorded into our audio file because we haven’t filtered it out.
And because the 80kHz wave is still periodic, it will record periodically into our 44.1kHz system, and we’ll end up getting an audible wave recorded!
This is called aliasing, and the way we talk about it is that the inaudible sound “reflects” back into the audible spectrum.
So how do we stop this? Well, we need to filter out those unwanted frequencies before we sample the audio. This is called an anti-aliasing filter.
Now, ideally, we cut literally everything above 20kHz, right? But in the analogue circuitry world we can’t do super super steep filters without causing some resonance - so we have to make some trade offs.
Some good audio manufacturers will be able to use better components and better designs to make these anti-aliasing filters more effective, but they’re still not going to be perfect!
So - what if we had more space before those aliasing “reflections” happen? Then we could use a gentler filter slope that would introduce less resonance!
This is the primary reason we have higher sample rates. Because a 192kHz sample rate can accurately record frequencies up to 96kHz, we have between 20kHz-96kHz to put our filter, rather than the 20kHz-44.1kHz of a 44.1kHz sample rate.
So, when people say “it doesn’t matter, you can’t even hear it”, they’re kind of right! We can’t hear above 20kHz, and our anti aliasing filters have got a lot better as technology improves.
However, the people saying “I can hear a difference”! Might also be right! If their equipments AA filters are better at higher sample rates, you get fewer “reflections” from frequencies above the sample rate, and that will sound cleaner.
It’s also possible that in some situations where some speakers can recreate higher frequencies, you might feel a difference - either humans are able to interact with sounds above what their ears can hear (I’ve not seen and studies or evidence of this), OR those extra frequencies are adding distortion in the speaker’s audible range (way more likely as speaker manufacturers are not checking for a flat frequency response above 20kHz [unless you’re Focal - I think their SM9’s are flat to 40kHz!!?])
So those are the facts! High sample rates serve a mechanical purpose, and it is possible that in some situations we can hear a difference between different sample rates.
My opinions - personally, I’ve not been able to hear a difference, and I haven’t been able to do a proper blind test to check!
You’d need two identical computers and interfaces playing at exactly the same time, switched through the same set of speakers in the same room - not many people have that!
The other disadvantages are that higher sample rates generate bigger files. This might not be a problem with storage as cheap as it is these days, but it also means more effort for your computer to process the same signal. So if you're getting CPU errors, it's not worth the potential gain!
I stick to 44.1kHz, or 48kHz if I’m working with video.
However, there is one useful reason to record at higher sample rates - if you’re doing serious pitch shifting and time stretching, your files will have more definition - you could slow down a 192kHz file to 1/4 speed and still have a real sample for every playback sample at 48kHz!
So there you have it - hopefully that's demystified the sample rate question for you, and if it hasn't, leave a comment and I'll do some more explaining!