There are several things to say about this. Firstly, hubs don't produce noise on mic signals! Wherever that's come from (either mic or mic preamp) it's 'analog' noise created by some analog electronics. You haven't said what mic or preamp (if there is one) you are using, so we can't be any clearer about that without more information.
As for your file - well that is actually a good example of why you sometimes have to go to extremes with NR when it comes to speech. Initially I couldn't get it to reduce anything but the introductory silence - but when I'd changed the FFT size to the largest there is, reduced the smoothing to around 200 and also the transition width down a lot I got some better results. You can't reduce the spectral decay too much, as artifacts become more prevalent. The best thing to do with any NR is not to do it in one go, but multiple passes, redoing the sample at each pass. And don't do them all at the same FFT size - use at least three - the lowest, highest and one in the middle. This always gives better results (even on iZotope's RX, which is what I use mostly for this), and is the best approach to minimising the bubbly noises that are seemingly unavoidable otherwise. And have a good go at the Advanced settings - they are the ones that can make a huge difference. And persevere...
2
Replies
AdChoices