Compressor study (windows VST)

moses

moses

hardliner
I'd like to present you the result of my latest compressor study.

Some of you already had the chance to try out a few of my creations, but they were shared privately. I think that it could be a great idea to share the result and discuss the details in public, so let's give it a try.

The whole point about this compressor project is that I'm not happy with the sound of most digital compressors. So I started experimenting with my own ideas several years ago. This is one of the several custom vsts I like to use in my own mix/mastering projects.

The compressor is of the "downward compression" type and uses a feed-back detection structure. It offers all classic controls: threshold, ratio, knee, sidechain hi-pass freq, attack, release, blend (dry/wet) and make-up gain. The controls should be familiar for most. I'd be happy to step further into details or design decisions, feel free to ask!

compressor.png



All you need is a windows PC and VST compatible DAW. Hit me up via PM and I'll send you the link to the VST plugin.

...oh, did I already mention its nice sound? ;D

download here: http://www.tokyodawn.net/tdr-feedback-compressor/
 

Attachments

  • compressor.jpg
    compressor.jpg
    692 bytes · Views: 234
Last edited:
You can't compete with Beetmaker VST, Mano is a genius. ;)

PM is sent..
 
Last edited:
It is a very nice sounding compressor - noisy on turning on/off by automating the vst host, but quiet when switching on/off in the vst itself.
 
I'd love to test out!

My gut feeling says you may know a thing or two about compressors lol. Sending a PM now
 
bandcoach said:
It is a very nice sounding compressor - noisy on turning on/off by automating the vst host, but quiet when switching on/off in the vst itself.
Click to expand...

Oh yes, the compressor introduces latency. 151 samples to be exact. The plugins "on/off" switch properly compensates the delay (the bypass is always delayed), but the "bypass" functions of most hosts might introduce a severe click. The host is basically informed about these 151 samples as it should, but it's left to the host to use it or not.

The main reason for the latency is that all relevant compression elements are oversampled 2 times. The oversampling is done via linear phase FIR filtering that delays the input.


For more details about the compressor architecture, here's a simplified overview (without the oversampling struture):
schema.png


For those who are less familiar with compressors, note that this compressor listens to its output. That is, the sidechain/detection circuit is connected to the output of the compressor. This structure isn't very common in digital compressors, most of all because they aren't trivial to implement. On the other hand, the feedback structure is very common in the analogue field.

The reason for the feed-back choice is that they have highly musical properties "by design": The ratio/threshold/knee and timing become dynamic, in a very nice way. But these compressors are less flexible, more difficult to compute properly and introduce far more distortion than feed-forward setups.
 
I am always interested in new pieces of gear so if you want hook me up with it and I'll try it out.

The screenshot looks real nice!
 
Thanks for the great feedback via PM! The plugin seems to work fine in Soundforge, Reaper, FL, Live & wavelab. Can anyone confirm more working DAWs?

A few more details (for all those who need that marketing yada ;) ):

- 64 bit / 2x oversampled differential sound engine for good signal path purity and solid processing quality.
- dynamic integration time RMS detection.
- instant feedback structure (not that easy in the digital world).
- optional auto release algorithm, dynamically adapts the release curve "shape" (not just its timing).
- super fast attack.
- typical feed-back compression sound with lots of tone and warmth.
- delay compensated "blend" feature.
- Optional low freq saturation feature (subtle and dirt free!) (not visible on the screenshot above).
- Optional Sidechain 6dB high-pass.
- Clever stereo linking network with cross-modulation suppression.
 
Last edited:
"Blend"

I've seen quite a few digital compressor and emulators but I've never seen any with a dry/wet control. Looks great! How can I get a copy?

and what's the range of the ratio values? Can it also be used as a limiter?
 
Yuno said:
"Blend"

I've seen quite a few digital compressors and emulators but I've never seen any with a dry/wet control. Looks great! How can I get a copy?

and what's the range of the ratio values? Can it also be used as a limiter?
Click to expand...

I sent you the link.

Well, there are several reasons for the blend feature. First of all, it's a feature that I use all the time. It has a very useful effect, especially in mastering applications (it effectively "relaxes" the compression curve, it can move the compression focus to lower levels). Second, the compressor introduces latency. That's why i decided to take the dry/wet on-board to guarantee a proper latency compensation. Oh and, third, the dry/wet concept is a feature that actually reduces distortion. :)

The range of the ratio parameter is very limited. It can reach up to 20:1 under certain conditions and never more than ~25dB of reduction. This is a common behavior of feed-back compressors. They see what "they've done", which makes it impossible for them to accurately reduce the level. Note that the ratio is extremely dynamic, much more than the static curve you can tweak via the "knee" parameter. This is the good side of feed-back compression. :)

The compressor is a very bad brick-wall limiter. The detection circuit not only listens to the output, it's also RMS based. The compressor doesn't even "know" about the exact input level - which makes it impossible for him to control overloads with technical accuracy. But you can still compress the signal like crazy, very much like analogue limiters do. But don't expect another L2 ;)
 
Last edited:
man after testing it out, it's solid. Seriously good stuff. I think another reason I can already think of so many uses is the fact you laid everything out about it.

I tried it on some vocals from a remix I did for Lights and she sounds good :D
 
It works nicely. Very smooth sound. Hard to clip it and make it sound harsh. Not an overly aggressive knee, which I like, since I rarely go for a really hard knee with compressor. The blend feature is handy. No funky quirks like some waves compressors, though I think waves does it by design. Transparent sound, which is good for the most part.

It's more along the lines of rcomp, which I like, because it's a no-nonsense approach to a compressor. A little bit smoother than rcomp and rcomp has something going on that I can't explain but I tend to like.

As far as using this a limiter, it will let some of the first peaks go through when you go for extreme compression so it's not what I would use it for. But it's not intended for that and that's typical of compressors so it's not a negative, it's just how it is.
 
@sleepy: yes, I like the rcomp comparison. :) I'm a big fan of waves' clean interfaces and great sound. they avoid any kind of questionable voodoo and still sound great.

@Yuno: thanks for the feedback, glad found some use for it! :)

yes, the compressor will most of all even out the RMS level, but it won't control the fastest peaks with accuracy. A limiter is perfect to technically even out the peaks, a compressor is usually designed around the idea to even out the RMS region and let transients pass through. In some ways, limiters compress the waveform, compressors compress the music. They do it with similar techniques, but the motivation is totally different.
 
Last edited:
Great job! It's very nice sounding. I test it as buss compressor, for which it's almost perfect because it's smooth. I love the "wave shaper" on the bass and HPF on the sidechain which avoid the usual damage on the bass impact produced by usual compressors.
But I would prefer to have the make-up gain only on the compressor section before the mix control.

My 2 cents ;-)
 
Thanks Laurent! The Low Saturation is basically just a maxxbass style processor, it's not related to the compressor algorithm. But I also feel that it can gently compensate the subjective loss or bass, especially on the mix bus.

Your idea about the mix/blend after gain make-up is interesting.

So, I currently do:
( wetInput + dryInput ) * makeup

And you mean this one, right?
( wetInput * makeup ) + dryInput

The latter looks like we need an additional output knob, but it could have a more natural handling. I need to try this out in the next days.
 
Last edited:
You must log in or register to reply here.
Back
Top