different wav bit rates in 32/64 bit apps - how to preserve the best sound quality?

[Oxytoxine]

Hi you mastering masters;

I am quite unskilled in technical questions, so maybe some of you could help me with the following:
When I render vsti instruments to wav files and then reimport them to work further with the wav files, what bit depth should I choose to get minimal loss of quality if the final product will be 44.1/16 bit cd ? Does it make a difference wheter I first “freeze” the vsti tracks as 16 bit, 24 bit or 32 bit files? I then reimport the audio to work further with (very convenient to comp e.g. improvised piano tracks or to work with stylus rmx) and after the mastering mix down the whole projects to 16 bit files. I do not really get how this conversions take place (and of course do not expect an answer to this, just a simple summarizing piece of advice would be very nice), I assume it also dependes on in which host one is working (samplitude 32 bit float, reaper 64 bit)?I normally record the rest of my instruments / vocals 44.1/24 bit, and i don't know how it interferes when i then do different for the vstis.
I hope the question was clear and not too stupid, otherwise forgive me, I’m in the first line musician not engineer.. I’m not sure wheter I really can hear the differences (unless of course if I add effects like delay or reverbs etc already in the initial step) or wheter it’s a “psychoacustic” effect, so to know what maths / physics say would be comforting, I just would like to prevent unnecessary conversions to get the best possible results.

Thanks for sharing you expertise, it’s very nice to learn from people who know what they are saying (not like some guys in the local music store)!

[mal]

Stay at the highest bit rate you can...if you need to render down to a lower bit rate then use dither if you can ..eg: from 32 to 24...but avoid going down to 16/44.1 ...most mastering places will be able to use their DAWs to process/replay your higher bit files...

[Kiwiburger]

If you record with 24 bit converters (which you should) then you never want to downsample to 16, except for the very final mastering step. That's assuming you are mastering your own work (which you shouldn't). So stay at least 24 bits all the way.

Exporting from a DAW with 32 float or 64 bits internal is actually a very good question. If any processing has been applied at all (includer fader adjustment) then it's probably wise to export at 32 or 64, to avoid further rounding error losses.

However - in a 64 bit DAW, that's going to generate a lot of filespace - it's a very good question.

My current thinking is that we should treat our DAWs as though they are fixed bit depth DAWs, and never get anywhere near 0dbFS. So the theoretical advantage of a floating point (being able to exceed 0dbFS) should never become a real issue (except inside plugins or the mix bus).

So my best guess is to export everything at 24 bits fixed, knowing that it will never clip because you aren't pushing your luck.

But whether there is any audible loss, especially in a 64 bit DAW - I don't know, and would be interested in any real guru's opinions.

(Admitttedly this is more of a mixing concern than mastering, but it would affect stems. Plus mastering engineers seem to be more onto this stuff that mix engineers).

[Oxytoxine]

thank you for the replies guys!

kiwiburger, from your number of posts i would well judge you to be a "guru"!;-)
i'll stay at 24 then and in the next holidays try to do some "in depht" research in how this all is interlinked. it's amazing, one starts with the simple attempt to record some piano tracks and ends with reading about nyquist equations etc... should better have become mathematician than molecular biologist..
have a nice weekend and thanks again!

oxy