16 or 24 bit

[Juho L]

The audible differences between 24/96 and 16/44,1 mixes are really hard (or impossible) to notice without really highend system (and without really trained hearing), so 99,9% of the listeners won't recognise 24/96 and 16/44,1 from each other. The main reason for using higher bitdepths is mixing. With more bits you get better resolution on low signal levels and this enchances the mixing process. If I'm not completely wrong every main sequencer software processes the audiofiles as 32-bit floats. This means that the waves are mixed in 32-bit resolution no matter what bitdepth the files actually have.

But I'd advice for going 24/96 just in case, eventhough it might be possible that no-one won't hear the difference. Better safe than sorry. The only situation where I notice bitdepth reduction is when I convert my project to a wave file (that annoying it-sounds-better-on-the-sequencer effect, which is caused by that 32-bit internal processing).