Downsampling and Dithering procedure.

[electro]

Which comes first when downsampling to 44.1 before CD burn.

Limiting or Downsampling? I have heard conflicting views.

This is how I have it.

1.Downsample to 44.1Khz
2.Brickwall Limiting @ 44.1khz
3.Dither to 16bit

[Aleksey Vaneev]

From my developer experience, this is right. Of course, given the brickwall limiting is used softly, peferrably with built-in oversampling (like in Voxengo Elephant).

[Kingston]

I know you want absolutely perfect math precision. Hence do all processing at as high sample rate as possible. That includes brickwall/mastering limiting as well, as it's marginally better sound. You need to leave about -0.3db headroom there for various CD player related compatibility reasons.

Then downsample, then dither.

Theoretically you can get intersample clipping when resampling brickwall limited material, but I've found this absolutely never happens, especially with that -0.3 headroom.

[Aleksey Vaneev]

Kingston, if we are talking 'analog' (minimum-phase) resampling like in r8brain PRO, it can bring a pretty high intersample overs into the sampled data. That's why 'limit after downsampling' is a more universal suggestion.

[Kingston]

Aha, good to know. thanks.

I've never used r8brain pro you see. The free one has been more than adequate this far. I think it's time to give it a spin.

What do you mean pretty high overs? more than a dB?

[Aleksey Vaneev]

Kingston, yes, it may give overs of around 1 dB - just like the actual playback equipment. Linear-phase filters are much better in this respect, of course.