FirComp: Free lookahead compressor for tracks or busses, looking for testers

[jonvaudio]

jonvaudio.com/fircomp/

FirComp is a free, clean compressor with low CPU usage, punchy attack, a highly program dependent and musical release stage, and an extremely straightforward GUI for fast workflow in a busy session. It also has optional FIR digital gain smoothing / lookahead. It's a deceptively simple plugin designed to sound good in almost any situation.

Testing versions of Fircomp are available as a 64 bit VST3 for Windows or Mac OSX, and a 64 bit AU2 for Mac OSX.

This plugin was my "lockdown" project: I've always wanted a highly optimized clean / lookahead comp that I could use on tracks or buses while not having to think too hard about what I was doing.

I have tested this plugin extensively myself, but there may be bugs or issues that I have not discovered yet. I am very happy to listen to feedback and bug reports

[perpetual3]

Looks excellent! About to test.

[bmanic]

Gave this a quick test.

It's release is way too program dependent, meaning the actual release control does almost nothing. Not much difference when set to 5ms versus 600ms. Maybe it's bugged? I also suggest giving it a much wider range for release, especially if you are going to keep it very program dependent. Then allow for at least 3 seconds of release time.. I'd even prefer 5 or 10s to truly let it work as a proper leveler.

Also, the attack sounds quite weak but this is usually the case with stuff that uses a lot of look-a-head. Do you have it properly setup with an actual hold for the look-a-head or does the internal sidechain actually trigger the attack before the main signal even arrives? This usually results in unnatural compression and a feeling of a weak attack stage at faster attack settings.

So in conclusion, while it does do "compression" as it is supposed to, it doesn't actually "sound" all that good to my ears. The release is well done but because I can't actually tweak it to the groove I have going, it makes this particular drum beat I was using less groovy instead of more groovy that I was going for.

Remember that compressors are far from just peak taming devices.. they are transient shaping groove machines.

[jonvaudio]

Gave this a quick test.

It's release is way too program dependent, meaning the actual release control does almost nothing. Not much difference when set to 5ms versus 600ms. Maybe it's bugged? I also suggest giving it a much wider range for release, especially if you are going to keep it very program dependent. Then allow for at least 3 seconds of release time.. I'd even prefer 5 or 10s to truly let it work as a proper leveler.

Also, the attack sounds quite weak but this is usually the case with stuff that uses a lot of look-a-head. Do you have it properly setup with an actual hold for the look-a-head or does the internal sidechain actually trigger the attack before the main signal even arrives? This usually results in unnatural compression and a feeling of a weak attack stage at faster attack settings.

So in conclusion, while it does do "compression" as it is supposed to, it doesn't actually "sound" all that good to my ears. The release is well done but because I can't actually tweak it to the groove I have going, it makes this particular drum beat I was using less groovy instead of more groovy that I was going for.

Remember that compressors are far from just peak taming devices.. they are transient shaping groove machines.

Thanks for taking the time to test this out and provide feedback.

Regarding the release stage, this may be a bug. Here is a quick & dirty screenshot of 10ms vs 300ms release curves on the same rectangular test signal - this is in the linear domain (I can generate a nicer excel test graph tomorrow with dB values) but it shows that the release time is supposed to change a lot. However, I would happily add much longer release times if needed. Please also bear in mind that the gain reduction meter itself can be slower to return to 0 than the compressor:



Regarding the attack, it is the case that attack with lookahead will be softer, but it can be punchy with faster settings and more aggressive compression (try attack times of 4-8ms and a lot of GR). Here is a dry vs wet example on drum samples. However of course this is subjective, and lookahead does smooth the transients somewhat.

I would also like to add that this compressor was not just designed to be a peak taming device I mostly use it with non-zero attack times to emphasise transients. If you let me know what platform you are on, I would happily send you a version with a wider release range at some point tomorrow. You have also made me consider that perhaps shorter lookahead times might be preferable.

[bmanic]

There are many ways of doing look-a-head but in general I've not really found any truly "musical" applications for it yet. It's always used/needed for transient troubleshooting (that is, catching them or making sure you don't get distortion when used with short attack times and release times, for instance kick).

I'm pretty sure I've got a bug with the release as I can really not hear a big difference at all when simply compressing a kick. I expected it to at least lift up the tail of a kick drum that had some room ambience in it with a 5ms release but there is none.. sounds virtually same at 600ms release.

I'm on Windows 8.1 running Reaper latest version.

[jonvaudio]

I'm pretty sure I've got a bug with the release as I can really not hear a big difference at all when simply compressing a kick. I expected it to at least lift up the tail of a kick drum that had some room ambience in it with a 5ms release but there is none.. sounds virtually same at 600ms release.

Ahh, now I am starting to understand where you are coming from Yeah, fircomp won't be good at that. It can make a kick punchy with an attack time of 7ms and other aggressive settings, but it's not going to bring up tail detail in a solo kick track.

Fircomp is more for situations like "I have a slightly distorted vocal from a client, I need to compress it a bit without adding more distortion". Or, when you already have the sound that you want, but you need to "control" it.

It's also for situations where you have something like piano or acoustic guitar, and you really want to crush it so it can be at the front of the mix, but you don't want any distortion.

However, there is one way to check for certain whether the release is bugged or not, and that is to duplicate the track, flip the phase, and see if they null when you change the release time on one of the tracks. If you could do this test for me I would really appreciate it

[bmanic]

Update: The version I downloaded is not bugged. I was just trying it on individual tracks and most of them trigger the program dependent release so hard that they always end up kind of slow(ish). So the 5ms release ends up being quite close to the 600ms release. On full program material or more harmonically and dynamically varied stuff I can definitely hear the difference.

I just can't put my finger on it but something is a bit "uninspiring" about the action of this compressor. I'm not entirely sure if it's the attack or release, or combination of both.. or the look-a-head. It does however work well as a tool to keep dynamics in check, however it has some issues. The way the transients get forcefully controlled and everything becomes slightly "dead" and neutered bothers me.

Here is a quick comparison on a little piano snippet where I do about 7dB of gain reduction. I tried to match the highest peak on both FirComp and my own algorithm. Note the initial transient and the immediate release compared to the original. FirComp gets quite unnatural in the way it squashes everything together, when it should be possible to do both squashing and keep it natural. My algorithm does however have a quite sophisticated release (multiple stages and sort of some feedback in the whole thing) but the main part is the attack.. in this example I used a simple 0.1ms attack time without any look-a-head. I could not match it in any way with FirComp, no matter how hard I tried. It either kills the initial transient too much, altering the macro dynamics of the chords or it sort of "slaps down" on the transient in an unpleasant way if I open up the attack. I was not able to maintain the original fluidity of the chord + immediate sustain part.

FirComp was set to soft knee. My compressor was set to about 12dB knee (half under threshold, half over threshold) so that constitutes soft knee I think. Both were 2:1 ratio. Both had attack at 0.1ms (though I did try different attack times for FirComp) and both have release at 600ms which obviously doesn't match even close but they do have some similarities.


Piano clip, original source

Piano clip, FirComp

Piano clip, bM comp experiment


This example is of course extremely brutal (it's why I use this little snippet as one of my go-to audio test files) so don't get discouraged!

What I'm trying to get at: Compressors are extremely cool in the way that it's virtually impossible to make one that sounds good simply by thinking "least distortion, mathematically sound!". At least that's been my experience and I've been at this quite a while now!

[bmanic]

.. oh and just for fun, here are both set to their fastest release possible. FirComp says 5ms and mine about 12ms but of course neither really does that fast except for maybe the first fraction of a dB.

Both compressors retained all their other settings from the previous example.

FirComp, fastest release possible

bM comp experiment, fastest release possible

Mine is definitely clearly distorting quite heavily (but it is much faster) while FirComp remains clean.

[jonvaudio]

I just can't put my finger on it but something is a bit "uninspiring" about the action of this compressor.

This was actually the design objective for the compressor! However I do really appreciate the time you have put into testing it even though you don't like it And your feedback has been really useful.

I have listened to your piano, and I get where you are coming from. I might take some of your suggestions on board esp. regarding what you don't like about the attack stage and how it interacts with the release stage.

Some background: when I produce music, I usually use Ableton's built in track compressor, or The Glue (both great). I often record instrumentals through a hardware DBX 160a. But fircomp is more designed for when the sound design / production has already been done, and you just want to clamp something or smash it to pieces so it can stand at the front of a mix without being fundamentally changed.

However, if I were to post a version of fircomp with a different attack style and/or a slightly more "coupled" version of attack/release topology, and with lookahead disabled or much shorter, would you be interested in testing it? Would probably be later today or tomorrow.

[jonvaudio]

.. oh and just for fun, here are both set to their fastest release possible. FirComp says 5ms and mine about 12ms but of course neither really does that fast except for maybe the first fraction of a dB.

Both compressors retained all their other settings from the previous example.

FirComp, fastest release possible

bM comp experiment, fastest release possible

Mine is definitely clearly distorting quite heavily (but it is much faster) while FirComp remains clean.

Here is a new test version of fircomp: https://jonvaudio.com/fircomp/software/ ... irComp.zip

You will need to re-scan the plugin as it has a new parameter. This version has:

- No lookahead (temporarily! )

- Different attack curve

- Attack and release are now "coupled"

- There is now a program dependent release percentage slider. I have not yet added this to the GUI, but you can access it from the plugin's generic GUI (screenshot from reaper):



However the release is still dual stage, even with no program dep. Am curious to know what you think of this.

[bmanic]

Ok, couldn't resist testing this even though I'm quite busy for the day.. but that's what lunch breaks are for, right? Testing new plugins!

The test version with the new attack is definitely a lot more pleasing sounding in my opinion. It's overall behavior is much more in line what I'm used to, thus it's easier to operate. Clicky tracks will still sound clicky now, whereas before the click would be sort of smeared and diminished. Now you can retain the click while it still remains pretty much distortion free. I also like how the attack knob/shape behaves now. It's much less "slappy" than it was before. Feels more natural in preserving the transient you want to preserve. Especially noticeable in the 1 to 3ms range where the previous version could get quite slappy.

I actually do like your program dependent release and seem to prefer it set to 100% most of the time but it's indeed a useful control to have available and gives more options in tuning the compression action.

Only had a few minutes with the test version so I'll get back to you on this later (probably over the weekend).

Cheers!
bM

[jonvaudio]

Ok, couldn't resist testing this even though I'm quite busy for the day.. but that's what lunch breaks are for, right? Testing new plugins!

The test version with the new attack is definitely a lot more pleasing sounding in my opinion. It's overall behavior is much more in line what I'm used to, thus it's easier to operate. Clicky tracks will still sound clicky now, whereas before the click would be sort of smeared and diminished. Now you can retain the click while it still remains pretty much distortion free. I also like how the attack knob/shape behaves now. It's much less "slappy" than it was before. Feels more natural in preserving the transient you want to preserve. Especially noticeable in the 1 to 3ms range where the previous version could get quite slappy.

I actually do like your program dependent release and seem to prefer it set to 100% most of the time but it's indeed a useful control to have available and gives more options in tuning the compression action.

Only had a few minutes with the test version so I'll get back to you on this later (probably over the weekend).

Cheers!
bM

Cheers, glad you like this version & thanks for finding time to test it. I have another version that you can test instead this weekend. It should be very similar to the previous, but:

- No lookahead, but improved envelope tracking to reduce distortion (noticably cleaner sound without any smearing hopefully)

- Attack curve very slightly sped up at the beginning of attack onset

- Release range changed to be in the range 20ms to 2500ms (meaning plugin needs to be rescanned again )

Link to file

[kingozrecords]

for a DAW like fl studio, look ahead is great, nice even in mixcraft as well, cakewalk maybe too amongst others. I'll check it out thanks.

[bmanic]

Sorry this took so long. Just downloaded the latest version and will give it a go right away! Got a bit sidetracked with another project and just spent a few hours lost in the magic of Wavesfactory's new delay plugin (it's amazing!).

[bmanic]

Quick first impressions of V3: Definitely like the attack of this thing more than the previous version. It's starting to sound pretty damn punchy now! Well done!

I also very much like the much longer release times. Now you can just "tickle" the mixbus with this thing set to moderate attack , long slow release with PD set to about 30% and you get awesome "mastering glue". Very nice!

This is a fine compressor now. Definitely recommend more people try it out and comment on it!

Oh and forgot to comment on just how awesome your preset sidechain filters are. I especially like the de-harsh filter type. Works very well as it says on the tin!