Cleanest most transparent vst compressor?

[tony tony chopper]

Its quite a simple concept really: a 3-band comp can be implemented with 2 dynamic shelving filters plus a gain stage.

Isn't it costy CPU-wise? I mean you have to update or interpolate your filter coefficients, while if you've splitted the bands it's easier to apply a smooth envelope.
The reason I wouldn't use them myself is also that I want to do other processing on bands, like saturation to start with.

[MeldaProduction]

I don't get the point. The thing is (which I'm pissed at), that Maximus doesn't keep the input frequency magnitudes the same even if the effect itself does not do anything.

It only does when you set a band very narrow, and even there the effect is less important than in the audio file I posted, and you know like me that the band isn't really narrow when you do that, it's just same but quieter, which is pointless.
Show me how a good 3-band division "doesn't keep magnitudes"?

but I'm offended if they claim

It is also a single-band compressor/limiter, thus can be transparent (what's your problem with that?), and can also be when set to linear-phase.
Does that mean that because your compressor features saturation as well, it's not transparent? I mean, it's there, so we have to use it, right?

A multiband compressor with non-linear phase filters can't be transparent, and yours certainly isn't. If I pass something through your compressor, that isn't compressed but sounds different, then it's not transparent. However you claim that yours is, and that it is in non-linear phase mode, that's the problem. The tone I posted only passed through your plugin, & you don't hear the difference?

It does that all the time. Only if it narrow you can hear that a lot. We are usually testing harmonic distortions and stuff like that to be around -170dB, here we are talking about up to -6dB.

If you want to make phase alteration considered nontransparent, then I guess we should stop using that word at all . Phase alteration is inevitable (unless linear-phase, where the artifacts are usually worse), but a crossover should be flat.

I don't understand the tone, it seems it is clipped (which means the compressor was clipping, not disabled), there is no original, so I cannot compare it to anything.

[MeldaProduction]

Its quite a simple concept really: a 3-band comp can be implemented with 2 dynamic shelving filters plus a gain stage.

Isn't it costy CPU-wise? I mean you have to update or interpolate your filter coefficients, while if you've splitted the bands it's easier to apply a smooth envelope.
The reason I wouldn't use them myself is also that I want to do other processing on bands, like saturation to start with.

Good point. Also most filters change phase alteration according to parameters, which means that the phase would be changing as the dynamic eq would work. It could matter sometimes.

[IIRs]

Isn't it costy CPU-wise?

My Synthmaker implementation is fast enough to be genuinely useful (uses about 2x the CPU of The Glue).

I'm sure a real developer could do much better...

[IIRs]

also that I want to do other processing on bands, like saturation to start with.

But we are talking about "cleanest and most transparent"

[IIRs]

Also most filters change phase alteration according to parameters, which means that the phase would be changing as the dynamic eq would work. It could matter sometimes.

I did wonder about that. So far I have yet to hear any problems however...

[MeldaProduction]

also that I want to do other processing on bands, like saturation to start with.

But we are talking about "cleanest and most transparent"

Yes, but you may really want to make it dirty and weird sometimes .
Basically I think the main trouble with dynamic eq used this way is that it doesn't provide much, it can really only eq, and it alters the phase. It usually doesn't matter, but on very low frequencies the phase alteration can start to be hearable, because it basically shifts the sound in time a lot, so it may be a trouble with bass drums for example. So the engineers may need to tweak the filter to make the bass drum sound fine, but it is impossible with dynamic eq, because the phase alteration is constantly changing.

[MeldaProduction]

Also most filters change phase alteration according to parameters, which means that the phase would be changing as the dynamic eq would work. It could matter sometimes.

I did wonder about that. So far I have yet to hear any problems however...

Yes, I usually don't see phase alteration problems as well. It can be measured, the waveform obviously changes, but usually it is fine. It may be a trouble with complicated signals, surround etc. though.

[tony tony chopper]

It does that all the time.

Show us all how the blank default preset does that? Maybe you just don't know how to use it?

If you want to make phase alteration considered nontransparent

Quoting yourself, "use your ears". If we use ours, your compressor isn't transparent.

I don't understand the tone, it seems it is clipped (which means the compressor was clipping, not disabled),

Just take a plain sawtooth, I'm sure you know how to generate one. And I'm sure you also already know it's very audible on such tones. The original is the first second of the clip.

[MeldaProduction]

So you are talking about the phase correct?

[tony tony chopper]

This is to point out the importance of phases. Each odd/even part in this clip is generated from the same partials, only the phases change, they're all aligned in the first parts, all random in second parts.

http://forum.image-line.com//files/phases2_102.wav

It's the phase relationship that's important, any same offset for all freqs won't be audible, but any abrupt change/discontinuity is audible (and I'd say that's where reverb comes from).
When you do resynthesis/stretching you really understand that phases (relationship) are a precious half of the info and that you're not gonna resynthesize anything without them (well you can, it will sound far/reverberated/phasey), and of course not preserve transients either.

If we get away with the dephasing in our crossovers, it's not because it's transparent, it's because it's usually not abrupt enough to be a big problem, & it's because it's often good musically (until abused), but not transparent.

Besides, it's one way to think of phases as something separate from freqs, but you can see the whole thing as modulating freqs, & thus by altering phases, you alter the freq content & magnitudes in a way.

[jens]

KVR at its best once again...

RedPhatt Pro from Jeroen Breebaart can be transparent or add some character/color depending on how you set it. Easily my favorite compressor.

Yes, Jeroen Breebart, just not RedPhatt - Jeroen has developed a different tool for that exact purpose - it's called 'PC2' and it works astoundingly well - but none of you guys managed to even mention it within the span of ten pages...

I bet if the thread title would have asked for 'the best compressor' or 'your favourite compressor' or 'the most analogue compressor' (or whatever) the suggestions would have been more or less the same.

[Dean Aka Nekro]

Voxengo Polysquasher is very clean

[MeldaProduction]

This is to point out the importance of phases. Each odd/even part in this clip is generated from the same partials, only the phases change, they're all aligned in the first parts, all random in second parts.

http://forum.image-line.com//files/phases2_102.wav

It's the phase relationship that's important, any same offset for all freqs won't be audible, but any abrupt change/discontinuity is audible (and I'd say that's where reverb comes from).

Hmm, can I ask how did you generate this? (I mean the algorithm, not the software). If such a thing could be caused just by filtering a sawtooth, that would be quite amazing for me. Anyway thanks, nice example!

I kind of suspect you used DFT for this, and in that case this would be possible, because of the discrete approach of DFT, which basically changes the magnitude content as well.

[tony tony chopper]

Hmm, can I ask how did you generate this?

Additive, bulk-sines, the initial phases are the only different factor.


I'm not reverb expert but I know that most are based on allpasses, I'd suspect they're the ones responsible for the reverb's diffusion quality, messing with phases.


It seems that the phase relationship between 2 partials stops mattering when they're over 1 to 1.5khz apart, but I'm no expert on this (& I don't know if that was ever studied seriously). So it seems to be a gap in Hz, not linked to critical bands then.