View Single Post
Old 10-16-2019, 03:21 PM   #10
serr
Human being with feelings
 
Join Date: Sep 2010
Posts: 12,562
Default

The recordings with the most damage have noise and coarse level swings that dwarf the actual program. Sometimes significantly. At the very least, you want to present the program louder than the background noise. If levels were moving around - go for consistent program level and let the noise and artifacts follow how they will.

Don't try to go for perfection with noise reduction. Go for attenuating the worst of it that's louder than the program. Be critical A/B'ing with the raw original as you go and don't sacrifice any content for trying to present it more perfect than possible. Divide the frequency range up and work different ranges with different amounts of noise reduction.

If you have resonant peaks going wild that would make the listener leap for the volume control, go after those. A little blurring at worst would be preferable to ice pick to the eardrum resonant peaks!

But yeah, if this is an mp3 of dialog only and already full of youtube artifacts... Put a low pass eq at 7k and call it a day.
serr is offline   Reply With Quote