Hello!

Nowadays I tend to always use RMS compressors because I mostly reach for a compressor to control the volume like an automatic fader.

If I need to limit peaks I use a limiter and a peak compressor only if I want to change the character of a sound.

My question is: why are most compression plugins peak?

I mean the very famous pro-c is just peak! (I have the official support message).

I often mistakenly hear that RMS compressors are nothing more than peak compressors with a longer attack, but nothing could be more wrong! it really changes the dynamics detection.

Thank you

 

More compression plugins are of the peak detection variety, because we have other plugins available to us which can do the "auto-fader" functionality you mention, without ever even reaching for a compressor plugin. Developers probably feel they have a better chance of selling a peak detection based plugin. If they offer an RMS detection option it's just extra functionality if a user wants it.

 

I think you're conflating two different things: how you like to use things and what things were made for.

RMS is average-based, so RMS compression is not useful in every case. In fact, I absolutely despise RMS compression, because if I reach for a compressor, I'm actually not looking to remove peaks, I just want to compress the signal after it hits a threshold (or drive into it). And honestly, I don't buy the whole threshold shtick, the "I pull up a compressor to catch peaks" or "I only compress if it gets *this* loud". I see no point in using a compressor if all you're doing is the same of moving a fader up and down. Just move the fader if you aren't looking to compress at all times.

If anything, as you're using compressors as some sort of half-useful gain rider, maybe just look into LUFS compression instead? Like the APU Loudness Compressor. At least it reacts to things in a more useful way for that purpose.

As for the question, "why are most compressors peak-based?", the simple answer for that is that peak compressors are simply more useful. The complicated answer is that dynamics processing in plugins is just code, and most code for dynamics (from compression to saturation) are the same thing with slight differences in implementation.

Maybe check:

1. Tone Projects Unisum
2. Woodstock Audio Open Compressor
3. Mixland Virelia

No one really needs more than the first two compressors, really. You can get the third one if you want to get fancy (and CPU expensive).

 

How do you like Open Comp? I was eyeing it when launched, but i have a lot of compressor plugins lol. Looks very useful though.

 

If it wasn't for Unisum, I would be using it a lot. I think the usefulness of it comes from the fact that it doesn't hide from you that dynamics processing is really very similar, so you get harmonics, limiter, upwards compression and other stuff bundled in because, well, most of the work was already done. The interface isn't great, but it's a very competent compressor where you can toggle everything. It seems that a V2 update is on the way with even more stuff, including mimicking other types of compressors like Unisum does. If I had to describe it, it reminds me a lot of those electrician pliers that can do 15 things in one package, you don't need to use all 15 things, but it's there and it's easy to use.

 

I am a friend of the definition for punch being a transient based signal, that has a lot of energy. Energy is defined via loudness over time. An RMS compressor does not add energy over time, a peak compressor does.

So the specification is: You take a peak based compressor if you want to add loudness over time. Which is the reason why they were used back in the time for this gain staging. Because the signal processors back in the days were just really bad.

There are people saying that punch is defined via the transient strength. I do not like it, because a cracking branch would be punchy then. This doesn't really make sense to me, because then you could say: "Yes, I do not want punch here, so I use a peak compressor".

 

Last edited: Dec 24, 2025 at 11:51 AM

Most compressors are peak because it's the most basic envelope detection design, both in the analog and digital domain, and is very versatile. They're most useful when trying to shape transients, by catching peaks or taming down the sustain etc...

RMS compressors are similar to analog opto compressors, in a way that they take an average of the input level over a certain window of time. This allows for a much more relaxed type of compression. It's better at equalizing the dynamics range of a performance, rather than shaping transients.

However, there is one more style of envelope detection that is to me the absolute most underrated technique in audio : the Hilbert transform envelope detector. Black Rooster KH-COMP and ToneBoosters Compressor 4 (in certain modes only) are some of the very few plugins using this. It works by extracting the envelope of a sound like this :


This allows the compressor to level dynamics in the most natural sounding way. RMS compressors, by design, need a certain delay to react to the sound. This doesn't, it can react very quickly. It's not good at shaping transients, peak compressors are still better for that. But when you want to compress the dynamic range, it really shines. ToneBoosters Compressor is my go-to vocal and bass comp. Its "complex" mode, a multi-band hilbert envelope detector, is incredible for mastering.

LUFS compressors are just RMS compressors with a K-weighting curve on the input sidechain.

 

Yeah got your point!

Back when I first started, I understood compression as 'a man moving a fader to maintain the volume.' But for a long time, I was mistaken because most compressors I used were Peak-based, whereas I actually needed RMS.

With a Peak compressor, the dynamics are controlled but often feel inconsistent—sometimes the sound is heavily compressed yet still feels louder in certain spots. Switching to RMS fixed this, as it provides much more stable leveling

 

I get you, LUFS compression was def a tide turner to me, haha. Fade riding is also important in that regard. Nowadays I'm mostly using compression for color, and I wouldn't underestimate Pro-C 2 if I were you.

 

My advice is trivial but anyway. May be what you are searching for is serial compression. One compressor is taming peaks only, another is affecting a signal more deeply.

It is more tricky to use but in some cases it can be of value.

 

All plugin compressors even the analog modelled ones are Feed Forward or hack job Feed Forward made to act like a Feedback comp. You can't do full Feedback compression implementations ITB as you cannot do instantaneous feedback loops without introducing delays ITB. Feed forward is very Peak topology friendly. Feedback requires that the internal sidechain see everything going on in the VCA, tube gain cell, opto, FET instantaneously. This is completely unfeasible in a plugin..plugins work from buffers

RMS has nothing to do with the envelope following contours and therefore not just a "long attack"RMS outside of feedback is difficult to implement. It requires squaring, lowpass filtering and integration of the internal sidechain to produce the correct sidebands(Nyquist limitations again)..none of these things are trivial to do hence why its not as common as you think it should be.

Very few plugin compressors can barely do peak compression right because of smoothing of the envelope follower. I don't even have the fingers or toes available to list the ones that do as most do(including Pro-C)and i can count the ones that don't on one hand

Personally, i would just use actual volume automation for what you are talking about. If your DAW can update volume automation at a sample level, then that is already infinitely more precise than some hack job attempt at an RMS or Feedback compressor plugin. If you are doing tiny amounts of gain reduction and distributing the compression across multiple tasks its probably not an issue, but if you are doing anything over 3-4dB it's probably introducing side effects of the design that you'd wanna avoid. Do the bigger volume moves with automation, massage with comps, saturation, limiting depending on the input

 

You don't need a compressor, you need a Volume Rider / Auto Fader.
Totally different beast.

https://www.waves.com/plugins/vocal-rider

 

This is not true. You said it yourself, a feedback loop is instantaneous, and therefore can be calculated without adding any delay ITB. Plugins indeed work with buffers, but the discrete calculation is done over a continuous discrete signal.

There is no inherent "smoothing of the envelope follower" in the digital domain, at least given the same bandwidth. It's a common misconception of the Nyquist theorem. The timing accuracy of digital audio is the same as analog audio given the same bandwidth and signal to noise ratio. With 32 or 64 bits float processing, you have way more than enough timing accuracy. You just need the proper discrete equations and/or oversampling to limit the quantization errors of the envelope calculation. But that's just a question of implementation, not a limitation of the system.