photonicrecords wrote:Sorry, lots of questions.
Yes, but now we're getting somewhere
... I hope I can present you with a logical answer.
Naturally, the output of an effect depends on what you put in -- but if this was the only requisite for the process to be called material dependent, pretty much anything could be said to be that. Instead, consider that there are effects which actively adapt their response to the material they're processing. This is important: some effects just "sit there" and constantly effect a signal according to certain settings (for example, typical reverbs and delays), but some effects adjust their behavior depending on what they're hearing (dynamics processors, autofilters and so on). Before getting carried away with that, let's sidetrack and recognize the difference between an insert effect and a send effect:
A send effect is not the opposite of being program dependent. When categorizing effects based on their placement in the audio chain, the fundamental usage types are insert and send effects. Insert effects are placed right in line with the incoming dry signal, and they output a modified signal back onto the same channel strip. In Live, when you want to shape a signal using an effect as an insert, you place the plugin directly on the channel strip of the corresponding audio source. Think of an EQ, for example: if you wish to equalize the output of a VSTi, you place an EQ plugin in the chain after the instrument. The EQ modifies the signal (and that signal only) and passes it down that chain, not letting any of the original signal bypass the equalization process.
In comparison, when using an effect as a send, it's used parallel to the dry signal. Instead of outputting a single modified signal down the chain, the ratio between the dry and the wet signal can be determined by the user. Most often this is done by utilizing the sends in the mixing console: the effect is receiving a signal through the console's effect sends, into which the user can direct any amount of any given channel's signal by adjusting the send level controls of the said channel. In this case, the effect itself is set to output only the wet signal (i.e. if it has a dry/wet control, this ratio is set to 100% wet, thus making it output only the resulting effect), and this wet signal returns to a mixer channel referred to as a "return channel." This is the way it works in Live, too: you just place the effect right onto a return channel strip, and by adjusting the corresponding send levels on the mixer channels, you can route multiple signals in various amounts into that one effect.
A regular reverb is the classic example of a send effect. Instead of using a new reverb on every channel you need reverb on, it makes sense to only use one instance of any given kind of reverb and send signals into it from multiple sources, at appropriate amounts, and have the wet signal return to the mix.
How does this relate to the sidechaining issue at all? We are getting there. Think of a typical send effect that just "sits there", receiving material from multiple channels and effecting it all, returning the wet signal to the mix. If the effect adapted actively to the material it's receiving, the overall result would be different depending on the combined input. In the case of a reverb, again, you would need to ponder whether the piano you're sending into it will still have the same kind of reverb when you send in some strings, too. Let's send a bit of that marimba as well -- can we trust the piano still having the same kind of reverb? Yes, you don't have anything to worry about, because the reverb doesn't "judge" the combined signal any different. A plain reverb just uses the same effect parameters regardless of the signal, and outputs the specified type of reverberation without actively adapting its parameters to the material.
Compare that to a dynamics processor. If you feed a compressor some bass guitar, you end up with a dynamics adjusted bass guitar sound. Now, feed it some drums along with it. When the compressor processes this combined signal, it's a given that the bass guitar reacts differently as well: as the compressor is a device that reacts actively to the input material, its whole output will sound different depending on the qualities of that combined signal. More specifically, a plain compressor is merely a tool for adjusting the level of a signal. If it attenuated the input signal by a constant amount, all the time, the effect would just make everything equally quieter. It would be no different than turning the level down -- in Live, using the Utility plugin to lower the level, for example. Instead, the compressor adjusts the level by listening to the signal amplitude and reacting accordingly. If the material exceeds the threshold level, it will be attenuated by a certain ratio, following the specified response curve. Similarly, a dynamic EQ effect doesn't modify the signal according to a static EQ curve, but the user can specify the conditions under which equalization should take place at a specific frequency range.
Take yet another example, a basic autofilter. Feed it a guitar, and you'll get a groovy (or not) wah-like sound if the parameters are suitable. The filter controls the cutoff frequency based on the input amplitude and different response settings, the specifics of which again vary from model to model. If you feed it two guitars at the same time, or a guitar and some drums, or the whole mix (now THAT'S a thing to autofilter
), the more complex amplitude changes of the combined signal throw off the nice accentuating filtering you had going with just the solo guitar.
So, on to the sidechaining debacle. A sidechain key input on a compressor allows the unit to react according to a different signal than the actual material it's attenuating. Because the device already contains the functionality to follow amplitude changes, it makes sense to modify the unit in such a manner that you can in fact "fool" it into adjusting the dynamics of a signal based on the amplitude of another. In addition to the ducking effect where a signal dodges when ever another signal (being fed into the key input) is present, another use is keying the compressor with an equalized version of the signal being compressed. This way you can isolate the specific frequency range the compressor will react upon, and this is what classic de-essing is based on.
A sidechain input on an autofilter can essentially be used in the same manner: as the effect already knows how to react to the input material, it makes sense to modify it into reaceiving another key signal than the one it's filtering. Routing a drum track as the key signal and a synth line as the signal being effected, the result would be a synth line which is being filtered rhythmically, the filter reacting to the amplitude changes of the said drum track.
At this point, let's spot the effects specifically in Live which actively respond to the signal being fed into them: at least both of the compressors, the gate, the dynamic tube effect (its tonal changes depend on the input signal) and the autofilter. I might well be forgetting something, but I hope the idea is clear. The Beat Repeat mentioned earlier was just an example of an effect that would be quite interesting with material dependent functions, as you could for example trigger repeats with amplitude changes and then add a key input for that.
But now, take the classic send effect, a plain reverb. An effect that just "sits there", effecting the combined signal of everything you want to hear that particular reverb on. Regarding sidechaining, the main question now is: what kind of functionality does a reverb already contain that benefits from adding a key input? Where exactly would the key signal go inside the reverb? What would it do?
Reflecting all of the above, consider these questions for a while. One can't say that an added key input would automatically make the reverberation sound only when there is key activity. That would be an active decision to add sidechainable gate functionality into the reverb itself. Perhaps a valid function, that, although it could be reproduced with a separate sidechainable gate -- and a gate like that would be more universally useful, as you could now make anything sound only when a suitable key signal is present.
Let's not make that judgement, however, and concentrate on the reverb. What is there, inside a plain reverb effect, that you could route a separate key signal into and make it do something? You're left with pretty limited choices. In simple terms, aside from being completely passed through, there is only one functional path for the signal to go inside the reverb: to go through the reverberation algorithm. Compare that to a typical compressor, in which you use the signal in two separate contexts: one, the signal actually being modified by the compression, and two, the signal being fed into the part providing the amplitude based control for the aforementioned process. So in this case, the unit's design already contains a ready-to-be-discovered idea for a separate key signal. "Hey, what if we feed a completely different signal into the control circuit, that would be cool!"
In the case of a reverb, it's important to acknowledge it would indeed need
some kind of added material dependent functionality for the key signal to be useful. There is no technical reason why a send effect couldn't have such functionality, so first you just need to decide what this dependency will be, and how will it be relevant in the context of that specific effect. Perhaps the reverb could change its frequency characteristics based on the input signal? That might be interesting. Or, will it modulate the room settings based on some attributes of the input signal? Only after you have designed what the active functionality will be and have successfully implemented it, then you can add a key input to make the unit react meaningfully to another signal than the one it's actually effecting.
... As it's a send effect, why aren't you just sending it the signals you need it to hear? If it's not even about the level of the wet signal, as you said, what does this "fooling the reverb into hearing a different track" accomplish, specifically?