Oversampling: How and Why

[malcolmwa]

My HiFiBerry Amp2 uses a Burr Brown TAS5756M. As I said in OP, I am trying to see what options I may want to make in Moode 7.0 Custom SoX recipes.

[swizzle]

I'm not sure what you mean by a DAC's "native rate". DAC's have a specific range of rates up to a max rate that they support.

You would need to look at the DAC's data sheet to see whether it offers an automatic, rate based filter bypass as is offered in the PCM5XXX chips. If not then its internal oversampling FIR filter will always be active.

What's the chip? I'll pull down the data sheet and see what it offers.

Resampling theory also posits that additional resampling done internally by the dac isn’t as critical (of course, but imagine upsampling an image from 100x100 to 1000x1000, if you use a very good method to go from 100x100 to 500x500 first then a simple one to go from 500x500 to 1000x1000 it’ll look better than if you just did the simple method all the way from 100x100). Here’s a good article that talks resampling with links to a few more prior on the subject.

http://archimago.blogspot.com/2018/11/musings-raspberry-pi-3-b-touch.html

Like with all audio questions you should listen to it and see if you think it sounds better or not.

[the_bertrum]

I'm not sure what you mean by a DAC's "native rate".

The rate to which the DAC upsamples with its internal interpolation filter, a rate which I assumed it would not bother to use its internal interpolation filters since the input was already at that rate. I realise this is a poor assumption based on the fact that the chip I have practical experience of (PCM5122) does have this bypass feature.

Like with all audio questions you should listen to it and see if you think it sounds better or not.

Absolutely, and a good thing too becasue everytime I think I've "got" the theory, I find I'm woefully wrong - but I know if I can hear something.
On the back of this discussion, I spent a few hours re-evaluating by sampling choices and I find that yes I do still prefer the sound I get when setting the sample rate to 384 kHz, but the bit depth has no discernable effect.

[bluefuzz]

everytime I think I've "got" the theory, I find I'm woefully wrong - but I know if I can hear something.

Actually, this is where you are probably wrong. When dealing with the subtleties of different (over)sampling frequencies and/or bit depths it is extremely unlikely you can actually hear any difference. Especially not without rigorous controls. Much more likely you are just fooling yourself. We all do it.

[the_bertrum]

everytime I think I've "got" the theory, I find I'm woefully wrong - but I know if I can hear something.

Actually, this is where you are probably wrong. When dealing with the subtleties of different (over)sampling frequencies and/or bit depths it is extremely unlikely you can actually hear any difference. Especially not without rigorous controls. Much more likely you are just fooling yourself. We all do it.

That is in fact exactly what I'm saying.  I can hear the difference between the upsampling done by SoX and the upsampling done by the chip in my DAC.  I can't tell the difference between different bit depths, or sampling recipies. That's why I suggested what I did in my post at 09-21-2020, 01:20 PM.  Malcolmwa can choose to learn about resampling if wishes for the intelectual challenge or he can try a few broad changes to see which he prefers, then get on with enjoying his music.