Kemper Amps Forum

Quoted from "MadH"

Quoted from "Per"

Yes that's what I meant (see my second post). And you may be right, but only Christoph can answer that. I'd be very surprised if the DSP couldn't handle it though given that even a Pod can do up to 96k and that's got a far less beefy DSP, also Christoph has said that internally it's working with much higher rates in certain places. But as mentioned, only Christoph knows if he can squeeze out that little bit more or not.

It would be sad if it's locked to 44.1 for it's lifetime, it would have been fun to use the Apogee convertors and keep an all digital signal path (not to mention less cables and therefore less hotswapping in order to use it), but being honest I don't hold much hope.

What I said is not the same as you mentioned on your post or at least not the way you wrote it. I am talking about using signals originally sampled at a higher rate, then summing and downsampling at the end. That normally produces better results but just because the source you are using is better/more accurate.

Maybe you were just confusing digital summing vs digital to
analog (mixer) summing. Some people prefer doing the summing in analog
even if all the source tracks are in digital because it sounds better for them.

No, that is what I wrote. i.e. Making the assumption (which as it turns out is wrong - see Christoph's post above yours) two approaches, high sample rate or reconstruction before modification of the sound at a higher rate (so effectively both = higher sample rate).

Anyhow based on the Nyquist rule (which is just a math rule, not specific to audio) then you'd need a sample rate of only above 2x the highest frequency that humans can hear in order to accurately reconstruct the waveform at the other end when accuracy = the same up to the limit of human hearing. So including what Christoph said above then unless higher frequencies/harmonics somehow were adding interference that's audible lower down (unlikely) then 44.1k should be just as "good" as 48k or 96k or even 192k, at least when it comes to human hearing, and the final result should null exactly if you were to take a mix made at 44.1k and one made at 192k and they were made to match frequency with the same aliasing filter as applied to the original sources for the 44.1k version (and according to what Christoph said that's regardless of whether they were reconstructed and re-aliased around the mixer/fx or if the samples went straight through the whole thing), and of course ignoring clock jitter. And I've heard that said before too elsewhere in these discussions.

I still prefer to work at 48k though, maybe it's purely psychosomatic, but it does feel easier to mix in and maintain clarity, or "better" as you say. Christoph has said in the past that there are parts of the KPA where the sampling rate is much higher, I wonder what they are and why they are there, also why audio software and interfaces offer higher samplerates at all, except maybe the cynic in me says as a meaningless checkbox to attract customers.

Quoted from "ckemper"

Oversampling two signals prior to summing and downsampling it afterwards will give the same results as when you just sum the two plain signals.
In other words: the glue that you produce by oversampling will be summed, but later discarded by downsampling, without having an effect to the residual signal. You don't gain anything by that.

Where the hell have your found this theory?
You will not find any skilled DSP programmer stating that summing signals is sort of a challenge.
Please tell me the source where this kind of digital esoterik is teached.

Digital summing is:
Have sample streams of the same sample rate.
You add every corresponding sample.
Especially when you mix many streams (mixing desk), maintain a high bit width until the last stream is added.
Then the result can be truncated (cut the bits not needed).

CK

Hi Christoph, thanks for chiming in. So I was wrong then with audio. Thanks for clearing that up. However I do think that this discussion is a diversion. As originally requested I'd still like 48k or other sample rates from the S/PDIF (or to allow the convertor to set the world clock/master rather than the Kemper), if nothing else I don't see it as in any way being a negative for the unit and it may even help bring in more sales (or at least not turn off anyone who may have been interested otherwise).