Compression and saturation, in what order?

[.jon]

True, and this info might actually be useful some day.

Time to go make some music before Finland beats Belarus on ice tonight

cheers,

.jon

[bobby yarrow]

The result of this thread is that some people who are just starting out are going to think it doesn't matter if they eq before or after their reverb sends, or if they put that space echo before or after their eq module. This is absolutely not the case. Every single recording engineer in the world for the history of recording can't be wrong about this, in everyday application anyhow. My stupid little demo shows an obvious difference, whatever the science predicts. And the engineer who is indifferent as to the order of effects is a guy I've never met.

I feel obliged to stress -- and I hope you'll join me -- that in application the order of effects is as important as gain structure or anything else.

[tahome]

I agree with bobby yarrow for two reasons: (1) In most cases you not only change the effects, you also change the routing. (2) Most effects have non-linear components. For example, an EQ might do saturation if you drive it with high levels, and a reverb might use some modulation to make it sound more natural. In all these situations the order of the effects actually does make a difference in practice, even though in theory it does not.

Hope that clears things up. If in doubt, trust your ears. I still find it interesting that in theory it doesn't make a difference in what order you do reverb, eq and delay.

--th

[Krakatau]

{edited}

[Music Engineer]

Well, lookit, while DSP technology is completely unknown to me, and while I don't doubt that a+b is the same as b+a, in all practical applications the order of effects like reverb, delay, and eq matter a lot. If you're describing a mathmatical claim about the properties of perfect processes in a theorhetical (but practically unobtainable) signal path, fine.

in the digital domain (and we are talking about this domain, right?) it is absolutely no problem to implement truly linear filters (including delays and reverbs) but i guess many devs incorporate some subtle nonlinearities even into effects which should actually be linear - maybe to add some "warmth", "character", "color" or other buzzwords or maybe to stabilize feedback-loops which would be unstable otherwise (which is what i did in my flanger and phaser). in the analog domain, strict linearity is more difficult to achieve....and...analog effects can produce noise - that obviously makes the position of EQ relevant. however, imho it is safe to assume that those inaccuracies can safely be neglected in the digital domain (at least when we use 64 bit processing)

[tahome]

Well, "abolutely no problem" wouldn't be my choice of words (it would imply infinite precision and resolution) but I guess what you are saying was basically my point.

--th

[Music Engineer]

Well, "abolutely no problem" wouldn't be my choice of words (it would imply infinite precision and resolution) but I guess what you are saying was basically my point.

--th

O.K., strictly speaking, roundoff is a kind of nonlinearity - but hey...in 64 bit precision ?

[Sascha Franck]

I always chain delays and reverbs. And I always put the delay first, because, regardless of what delay and/or reverb I'm using, the difference is huge.
Can you please tell me a single reverb/delay that I could try with where I wouldn't notice a difference?

Btw, you *might* be correct assuming a delay with a delay time longer than reverb time was used and in case the delays feedback is set to zero. In that case it won't make any difference. But as soon as these things are mixed up (more than one delay repeat, delay times rather short, reverb rather long), the difference is always there.

[tahome]

Again, in short (because I've explained this over and over earlier on in this thread):

Delay, reverb, EQ, filter are by their nature LINEAR effects. This means the property of additivity applies to them. Additivity means that adding two signals and sending them through the effect is the same as processing both signals by the effect separately and adding them together. This assumes infinite (or at least reasonably good) precision, and no non-linearities (like modulation, saturation and the like) in the effect chain.

By virtue of this property, it also does not make a difference in what order you process a chain of linear effects. In other words: GIVEN THE SAME ROUTING, it does not make a difference if you process your guitar with reverb and send that through a delay, or vice versa.

If the conditions mentioned above are met, you can plug your guitar into a reverb unit and plug that in a delay unit and listen to the signal. Or, you can plug your guitar into the delay and that into your reverb. It does not make a difference. None at all. That is what a linear system is all about.

In practice, the above conditions are rarely met. Most delays or reverbs use modulation (which introduces non-linearity) or add some kind of distortion (which is also a non-linearity).

If you find this difficult to believe try a simpler example first: Imagine two filters in series. One hi pass that lets all frequencies through that are above 1000 Hz, and one lo pass that lets all frequencies through that are below 5000 Hz. The output of this series of filters will be a signal which (roughly) contains only energy between 1000 and 5000Hz.

Now think about it: Does it make a difference whether the hi pass or low pass comes first? No, it does not. The result is exactly the same in both cases, because filtering is a LINEAR effect.

Is reverb and delay different from filtering? No. Reverb can be modeled as a (long) filter. This is in fact done in all the Impulse Response reverbs. Same for delay. So the same rules apply.

Hope that explains what I was referring to.
--th

[Krakatau]

I clearly knew the problem you're facing now , tahome, when i first came on this forum in may 2002 ( exactly 4 years by now ) trying to explain again and again some ideas that i was persuaded of their utility but couldn't really create continuous interest for them...

____


if you would ever be curious, i be glad to find again the links to the related threads ...

[tahome]

What kind of ideas?

--th

[Sascha Franck]

So, tahome, *any* single delay/reverb that would meet these scientific specs? Because I'd gladly try things out.

[tahome]

No need to try... but if you want to, any reverb based on a Schroeder/Moorer design would do the trick (Freeverb for example). Delay shouldn't be a big problem either, any reasonably clean (ie. non tape-simul) delay should do it.

--th

[bagginz]

Sorry Tahome, but you would have to be totally deaf to not hear the difference between a delayed reverb and a reverbed delay in practical setting (i.e music production)

Music production is a listening art. Theories of DSP coding are another subject entirely.

Cheers,
Bill

[tahome]

if you would ever be curious, i be glad to find again the links to the related threads ...

I am. Didn't I just say that?

--th