|
|
|
In a nutshell, you almost certainly can't remove reverb from a recording. Here's some exchanges in a recent thread:-
SteveG: You might think, if you looked at the 'convolution' effect that's available in Audition, that if you could convolve the sound of your room (create an impulse and record it), and instead of adding the convolved sound, subtract it instead, leaving just the sound of the original audio without the room.
Unfortunately, it sounds like a good idea, but it falls down quite significantly, for several reasons. Firstly, creating a suitable impulse in one part of the room is one thing, but you actually need a different one for every mic position. Secondly, if it's a small room, the effect of what's in it when the recording was made is significant, ie people, equipment, etc. and this would have a material effect on the quality of the impulse. And then you have all of the problems of level-matching the results.
What seems to happen about, oh, 100% of the time is that the resulting file sounds worse than the original, and damnation, there ain't nuttin' you can do 'bout it.
Ozpeter: You could try playing around with parametric (or other) eq, which might make allow you to emphasise the voice in contrast to the reverb which will have been affected by the room's acoustics, but you'd only be picking at the edges of the problem, as indicated above. Mis-applied noise reduction sometimes has the effect of reducing reverb but at risk of undesirable artifacts, but I've only tried that in respect of large acoustics.
SteveG: You could try EQ, but with respect, I'd say you were completely wasting your time! Unless by some very strange freak of nature, you've got a room that reflects back frequencies other than those that excited it, you'd only be EQing out those frequencies that you wanted to keep - and as a consequence be reducing the level of some of the wanted signal, which is going to make it harder to understand than it was already. So, I'm afraid that EQ is actually counter-productive.
One way that you can sometimes reduce the effects of a room (and this only works if you have a high direct to reflected sound ratio) is to use the dynamic processor to expand the (quieter) reflected sound downwards, so that it is reduced in volume relative to the wanted signal. But with most video productions, the recorded sound is all over the place, and it becomes virtually impossible to track the level changes. These have to be carefully controlled if this technique is to work, and it's no good thinking that you can compress the vocal to do this, because you will lose precisely that part of the signal dynamics that you want to differentiate.
In this situation, most people either give up and use it as a learning experience, or redub the vocals - a learning experience in itself if you've never done it before. But you'd be amazed at the results you can get, especially if you employ a good foley artist!
|
Author
