Input Aliasing

I was just mucking around with an oscillating filter being routed through the ER-301 and was rather surprised by the degree of aliasing artifacts being produced. I thought perhaps my filter was acting up, so I recorded some sweeps to get to the bottom of it:

The first sweep is direct and the second is being routed through an empty chain. The input level was moderate, so I don’t think the anti-aliasing filters are clipping. And the built-in sine oscillator does not exhibit any aliasing that is noticeable to me. The effect is more pronounced with the 48kHz firmware, but is also present with the 96kHz version.

I understand that the inputs are being resampled from 60kHz to the target rate. Is it maybe possible to run the IO at the ‘host’ rate to avoid resampling? Or maybe an option to run at 60kHz?

There is still some room for improvement in the input resampling algorithm. In the meantime, I can’t do 60kHz processing but I can do 62.5kHz processing:

er-301-v0.2.17-unstable-62.5kHz.zip (1.3 MB)

Sweet, thanks for the firmware. Operating at 62.5kHz doesn’t seem to have bought much, it still falls between 48kHz and 96kHz in terms of aliasing. If the resampling can’t be avoided then I will just have to wait until the algorithm is improved. Thanks.

Can you confirm that you are using the IN1-4 inputs?

Yeah, IN1.

Would this impact the looper at all, specifically when layers are recorded and dubbed? Sometimes I’ve noticed strange degradations when recording several layers of guitar. When I record with no layering and dub at 100, it seems good. Using input 1 and 96k.

I doubt it!

It would be much more likely that you have exceeded the maximum signal magnitude and you are clipping.

As far as I can tell it’s not clipping. Only happens when the old is blended with the new and sounds like aliasing. I could record and email some clips to you later when I get home, if you’d like.

Yes, I would like that. :bow:

Ok, will do when I get a chance, thanks!

We think completely differently, I see these audio quirks as features :smiley:

I am all for these things being ‘fixed’ but I think something is lost in the fixing process that is worth keeping and even developing and exposing as a parameter to be manipulated like anything else.

I know I am right in this (as in I am right that it is my opinion and view), but understand others disagree and that’s fair enough, but please, lets not lose this - it sounds great :slight_smile:

I’m quite fond of aliasing artifacts, quantization noise, etc. It’s easy to induce them if desired. What isn’t easy (or even possible) is to reverse these effects after they have been applied. I would prefer to work with a signal that is as close a representation of the original source as possible, with any additional treatments being applied to taste.

1 Like

I agree with this! Would be great to have the option to add this as an effect rather than it always being there.