Say I have a wave file, very low in volume, and with some constant background noise.
Is there any theory of what to do first: first de-noising it, then amplifying, or the other way round? I always thought de-noising should be done first, but I’m not so sure anymore, because you amplify the result of the de-noising process. Now I wonder if there is one way superiour to the other.
Seems I’m bored…
In most cases I amplify first, because the noise reduction plugin/algo precision is usually better then in my experience. I seem to get much worse artifacts when using noise reduction to low volume signal and then amplifying it than using noise reduction to already amplified material. Also it’s usually better not to reduce all noise because of artifacts, but take say 50% or 75% and handle the rest with gate.
I’ve also had couple of weird cases where I’ve gotten rid of very low volume noise in low volume signal with basically zero artifacts by just running it through ReaFIR in subtract mode and zero reduction. Probably something about the lack of precision in ReaFIR which results in noise disappearing.
Btw, I really recommend checking out Voxengo Redunoise. My new favourite noise reduction tool. Very very powerful and customizable and also works as a super equalizer 
Thanks sauli, I somehow suspected amplifying those artifacts might be a bad idea after all… And first spreading the values to some maximum might help, probably depending on denoising algorithm. Makes sense.
@Redunoise: Can’t afford it ATM. Will check it out when I can. 