When the mix is done, you want your song to be mastered. But how should you export your mix for the best result?
(tl;dr at the end)
Speaking about peak levels, theoretically your max peak could be -0.1 because what matter is to have a clean signal with no clipping, anyway it's suggested to go lower, and since there are no drawbacks to lower the volume have the max peak at -6db it's a good choice.
Of course peak level is not the only setting you must care, but sound quality matters too.
When you export your mix, it's suggested to use 32bit floating point if possible, otherwise 24bit is good enough, no worries. Anyway i really suggest you to avoid 16bit, cause there are too much informations that will be lost.
What about Sample Rate? You may find suggestions, articles that say the higher the better, while that's technically correct, i made lots of tests in the past, and in the real world this doesn't really matter 44,1khz it's enough.
While mixing, it's importhant thought, to use oversampling on every process that introduce harmonic distortion, so plugins such compressors, saturators and clippers need to work on higher internal sample rate (oversampling) to remove aliasing and getting back with a clean signal.
Oversampling make obsolete the need to run the whole project at higher sample rate, last but not least your music at the end of the process will be converted at 44.1khz, because even on the streming services era it's a standard, and by the way 44.1khz means to have 22khz as higher frequency. Kids can ear up to 20khz, while adults around 15khz (it's a bit more complicated, but let's keep it simple) so 44.1khz sample rate already allow us to have available all the music content human can hear.
I know nerds out there would say that it's still better to go higher sample rate because of Nyquist frequency.
If you filter at higher frequency the response on the 20khz may be flatter (it depend on the Nyquist filter and how the transition band is designed). And while this is technically true, very likely you will still need to downsample to 44.1k later, to distribute your music to streaming services (most of them don't accept anything different from 44.1k) and even if we can go higher, and we render with a very bad program that have a poor Nyquist filter, we would likely lose 2db at 21khz to 0,5 at 19khz. Can you really hear these differences?
I made my accurate tests in the past, the only difference in the analyzer were around -110db at the worst case, not really something we can feel.
Just a couple of suggestions:
I strongly suggest you to avoid any limiter on the mix bus, this may harm the mastering process.
Other kind of mixbus processing depends on your knowledge, if your song is mixed by a professional mixing engineer he know how and what to do, but if you are the one who mix your song, it's better ask to your mastering engineer what to keep and what to remove.
And here we come to my last suggestion:
When you hire an engineer, don't be afraid to make your questions, the more we talk, the better, don't worry, we don't eat artists and producers.... usually :)
Sample Rate: 44,1khz
Bit Rate: 32bit Floating Point
Max Peak: -6db
For any doubts talk to your mastering engineer
If you want your song to be mastered (or mixed) feel free to send me a mail at effettimusic[at]gmail.com