What setting do I use for my DAW with Kemper? 44100 or 48000 or higher?

    Good video, Alan. Looks like I've been sucked down the rabbit hole again.


    IMHO he didn't explain his preference for 48kHz over 44.1kHz very-well. He referred to the old-style analogue filters and common internet-delivery rates, but they're false arguments IMHO:


    1) The digital brick-wall filters used these days don't require the roughly-4k "width" 44.1kHz afforded the old analogue counterparts. Adding a further 4kHz cannot be justified in this way, therefore. Remember - 4kHz was already added to the theoretically-necessary 40kHz specified by Harry Nyquist in order to accommodate those filters' steepness limitations.


    2) Majority / popular opinion is not a valid argument.


    It'd be great if Christoph ckemper could take a few seconds to confirm my contention that 44.1kHz is in fact adequate due to the progress made in implementing this filter over the years. It's so-ridiculously steep now that it wouldn't surprise me if we could theoretically go back down to near the Nyquist frequency for sampling rates, IOW, less than 44.1kHz.

    Quote from Monkey_Man

    he didn't explain his preference for 48kHz over 44.1kHz very-well

    The "problem" is that the difference is pretty much negligable. Why?

    One octave up from 10kHz (roughly a D#9) is 20kHz (roughly a D#10), one octave even further up is 40kHz (roughly a D#11) already.


    This being said, if you look at 44.1kHz sample rate, the frequency overhead is 2050Hz (22050Hz - 20000Hz). This equals 169 cents or 1.69 semitones. At 48kHz the frequency overhead is 4kHz. This equals 316 cents or 3.16 semitones.

    Both being FAR away from 1200 cents (or 12 semitones or 1 octave).


    Now what does this mean? It doesn't make much of a difference considering that we typically speak of dB/octave. We have much less than an octave anyway. The rest is just a meta-discussion. There were technical reasons why 44.1kHz was picked in the beginning. And there were technical reasons why (at some point) 48kHz became more practical.


    The even higher sample rates like 96kHz or even 192kHz are yet another can of worms that I could get into more detail if asked for. But for now the above should at least show that you don't need to overthink this small difference between 44.1 and 48kHz. :)


    PS: And one more important thing I forgot to mention:

    If you apply a high-cut filter, the filter curve is a knee around the center frequency. The interesting effect is that the lower the filter "steepness", the more audible frequencies below the center frequency get affected. But the effect above the center frequency until the Nyquist frequency isn't all that much if you compare e.g. 12dB/octave and 36dB/octave. So to completely filter the high frequencies you'd have to use 48dB/octave or more anyway, no matter if you're on 44.1kHz or 48kHz.

    That's why it's a better option to use oversampling (and the corresponding filtering) in plugins that introduce high frequency harmonics (e.g. saturation).