Decorrelation. What's that all about?

[DG]

I've been playing with the Suite convolution reverb today. Very nice. However, I don't understand what Decorrelation does. I've read the manual, looked online, and I still don't understand. Is anyone here able to give a simple answer, preferably with words of two syllables or less, so that I can understand what this lovely slider does.

Thanks.

DG

[Karel]

In theory: It decreases cross-correlation (the amount of similarity between two signals) between every impulse response channel. Further reading: http://en.wikipedia.org/wiki/Decorrelation and http://en.wikipedia.org/wiki/Cross-correlation. In practice: It smudges the impulse responses to give you a more wide and lush tail.

[DG]

Thanks Karel. I did read the Wiki articles, but all they managed to do was give a more smudged understanding of what it was all about.

So just to clarify, for the clearest sound, leave decorrelation to 0, and for a wider, lusher (is that a word?) tail, increase decorrelation?

DG

[Karel]

Yes, but the decorrelation envelope also matters. If it's set to zero for the whole length of the impulse response, the slider won't have any effect. Just look at the included presets for inspiration and follow your ears ;).

[Dietz]

Hi Daryl,

The more correlated signal content L and R have, the more the result will seem to be "mono" - thus the term decorrelation. ... a somewhat nicer description than "smuding" would be "algorithmically enhanced phase difference between the left an the right signal for a more perceptible stereo impression and spatial enveloping". :-)

Kind regards,

[SJSF]

@DG said:

I did read the Wiki articles, but all they managed to do was give a more smudged understanding of what it was all about.

Decorrelation: the process or act of smudging one's understanding. lol

Thanks for the articles dietz, interesting.

-Sean

[DG]

I've been playing around with this slider for days now, and I find that it can give a much fatter, warmer sound. This has been especially useful when playing with samples from "another" company that are recorded in an anechoic chamber. After I've finished with them, you really wouldn't know that they weren't recorded in a big studio.

DG

[SJSF]

[quote=Karel]Further reading: 

Wow, I just realized my error. I meant, thanks for the articles to you Karel, and thanks for the very techy description to Dietz... That actually made this make more sense to me... surprizingly.

Anyway, I hate to pester over something so small... so I don't expect a reply... but does MIR have this feature? Does it need this feature?

-Sean

[Dietz]

Actually the feature was originally developed for MIR. It's part of the virtual "Distance" between the single microphone capsules of a Main Microphone array. (... "virtual" because the whole MIR concept fundamentally relies on Ambisonics, which is coincidence (read: non-spaced) setup by definition.)