Deverberation

[Svama]

Hi Vojtech,

is it possible to decrease reverb on signals with one of your plugs. MTransient is one way, but not the best! Other ideas for that task?
There are some deverberation plugs, but don´t like to buy more than necessary.

Thanks and greets

Marek

[MeldaProduction]

Hi Marek,

first of all, the existing deverberation solutions sucks imho. Unveil is usable, but it still sounds like sounds from Mars .

Anyway, I'd suggest trying to abuse some psychoacoustic laws:

1) You already found the transients.

2) Next, simple gating - just try MDynamics as expander, set the threshold pretty high as you want it to start working very soon after the main signal. The reverb will then be masked by the main signal, significantly higher in level.

3) Ideal reverb converges to white noise, so you may simply try denoising - MSpectralDynamics.

Technically you can never remove the reverb correctly. Reverb removal is even more problematic than noise removal, because unlike noise, which is persistent, reverb is changing according to the main signal, but evolves to white noise anyway. But using some of these tricks (even combined), you should be able to mask it well enough.

[Svama]

Will try, thanks a lot!

[CableChannel]

try denoising - MSpectralDynamics.

Wut? MSpectralDynamics can serve as a Denoiser? Why did nobody tell me?

Vojtech, your plugs have way too many hidden cool functionalities under the hood

[MeldaProduction]

Hehe yeah! Check the video tutorial

[nicocadiou]

Hi all,

interesting thoughs, but I don't see how working on transiant can achieve a kind of de-reverberation...

I thought it was about smoothing and flattening attacks, with a high resolution value, but I tried and it only leads me to bad artifacted results.
Maybe am I on a wrong way...

Wathever, I wonder if you can explain that a little more ? Or maybe do you have a web link about that.

Thanks

[MeldaProduction]

It doesn't remove the reverb! The point is to make the brain focus on the other part of the signal. Transient processing may only work with percussive signals. It would of course change the character of the "drums" as well.

[goldglob]

Vojtech was talking about expanding the transients, not compressing them.

[MeldaProduction]

Vojtech was talking about expanding the transients, not compressing them.

Good point!

[robertszalapski]

I agree about de-reverb solutions most aren't great. Even the demo samples to show off the product are pretty bad. The DSP would be non-intuitive and non-trivial, and likely need offline analysis. It would be awesome if someone figured it out, but I just ordered a ton of insulation - I'm going to make my room dead, acoustically.

[MeldaProduction]

Yep, that's the best way . The thing is, it probably could be done by sampling the room's impulse response first, but nobody wants to do that obviously and it would work if the sound source was coming from a single spot anyway, so well it probably wouldn't work very well .

[robertszalapski]

Exactly. A room's impulse response, even if it was 100% static (which it isn't quite), varies based on both position in the room and direction. The DSP to remove it would be slightly different depending on how the sound radiates through the room. I can't think of any way to remove generic ambience non-destructively, because how do you differentiate between the room ambience and the resonance of the instrument itself?

I'm sure unveil has saved people time and money by reducing apparent reverberation in recorded material, but it's a huge CPU hog and doesn't work well in removing small rooms.

I do think a better version could be done, theroretically. But saying it wouldn't be trivial would be an understatement. It's easy to get spoiled by all the digital tools we have and expect them to do more, and I hope we keep pushing the envelope.

But yeah, if you want good recordings, think about where you are recording and consider treatment options.

[MeldaProduction]

Cannot agree more...