[Bluelake]

Originally Posted by kjg

For anything serious, I upsample to 96k first thing. It is a drag, but it does sound a lot better. Compressors and limiters specifically.

If I need to get it loud the end of the chain might look something like this:
Sonnox with medium to slow attack, maybe a touch of knee, GR "leds" barely coming on > Elephant non oversampling (but running at 96k!), doing 1-3 dB GR with whatever settings are needed > ceiling to -.4.

Downsample > sequence, fades etc > kjaerhus mpl-1 pro with threshold at -.36 (it has an oversampling side chain, but leaves the audio alone), and unlinked up to 10-20 ms, to catch the stray peaks from downsampling > dither > export.

Normally I prefer to upsample once, let all plugins run at native high sample rates, then downsample once. Some plugins automatically upsample when you run at lower sample rates... Another good reason for me to do it just once before I start. You could end up with 3 times up and down in a complex session.

If I need to run Elephant at 44.1 I'll switch it to 2x or 4x oversampling, depending on what seems to match the program best.

I pretty much only like Sonnox for the fact that it has an attack time, and a knee. I never use it as a brickwall. I never use auto-gain (I can do my own macrodynamics, thank you very much), or safe mode (distorts too often, and generally colors too much). Sometimes I think I like 6 - 12 % of enhance, but I almost end up turning it off. I pretty much don't like the sound of it as soon as the lights come on consistently. But what it does just before that, just shaping the loudest peeks, that is nice. Just a slight tightening.

I'll be demoing Flux... I wonder how it compares to Xenon.
The variable linking in Xenon is nice btw! Too bad Flux doesn't have that.

Any Flux experiences? Especially compared to Xenon?

regards,
kjg

Thanks for this, very useful !
I was trying to find a way to avoid more than one downsampling, specially once the file is at 44,1 kh. Will be buying the Kjaerhus limiter that doesn't upsample and downsample the actual audio but just the sidechain....

Juan