Free SynthMaker Compressor+EQ/Multi-stage distortion unit

[egbert]

Question:
The auto-gain works as it should, for the sustained part of the sound. But when using longer attack times the overshoots easily result in distortion.

Is that because there is a limiter/saturation on the output or something? Make up gain doesn't usually cause problems for a track compressor - it can make the signal pretty hot though.

[stefancrs]

Question:
The auto-gain works as it should, for the sustained part of the sound. But when using longer attack times the overshoots easily result in distortion.

Is that because there is a limiter/saturation on the output or something? Make up gain doesn't usually cause problems for a track compressor - it can make the signal pretty hot though.

There is an optional saturator (the softclip knob controls it) but nevertheless, somewhere along the chain, even without the built in saturator, the overshoots (if being lots of dB above 0dB) easily ends up in distortion somewhere. If you pull down the attack to zero you don't have this problem since you basically enter the sustained part of the compressed signal immediately, which ends up at zero dB when using the current kind of automatic make-up gain.

It is really quite simple to take care of the peaks and make sure _they_ don't overshoot, but then the sustained part ends up below 0dB (far below if having a low threshold), which is why I thought a knob controlling the automatic gain compensation might be a good idea, but I really don't know and don't think I have seen a similar solution in any other compressor. All other compressors I have tried "suffers" from the same problem though, since it really is in the nature of compressors to behave in such a manner.

[egbert]

Compressors like the Kjaerhus ones or Voxengo Crunchessor or Marquis Comp will both put out signals way over zero (10 dB plus) if you turn up the output gain and this hot signal doesn't produce distortion. I think that is a desirable approach.

If there is to be limiting/saturation on the output - I would suggest that should be able to be defeated - eg Beatpuncher has a switchable limiter on the output.

[stefancrs]

It doesn't produce distortion here either if you pull down the softclip to zero. Like I said, the soft saturation is optional

[brok landers]

hey stefan,
i've read the whole thread, and i've been trying out your plugins.
first, let me say that these are really nice!
thank you so much for releasing them for free!
may i still mention some things, that could help to get to the state these plugs deserve:

eq:

- give the eq an input and a gain slider (the gainslider should be placed in frot of the saturation stage)
here is why i suggest the gain slider:
when eq-ing a signal, it can easily come to the situation that the level of the signal becomes loud, when boosting the bands. now, when i don't use the saturation, i don't have the possibility to readjust the output other than using the fader on the mixer. i don't like to use that mixerfader for this, as i need the full range to be able to ajust the signal precisely in the mix.
if i use the saturation, it could be that the signal gets way too hot, even if the saturation knob is set to zero, because of the signal being boosted way over the internal threshold of where the saturation starts to take place.
here is why i suggest the input slider:
some plugins are (like your's) not able to control the output, and sometimes they produce too high leveled output. placing your eq after it would set me to the need to lower the input before actually processing any further.

- numeric readout-dragging/editing.
here's why i want that:
i (for one) like to be able to edit rather via numbers than dragging the graphics, as this is too unprecise for me.
also, when eq-ing the very high content of the sound it could happen, that the q-node runs out of the display. to edit the q then i have to move the frequency node to a deeper point in the frequency just to adjust it ... that at a frequency that i don't wanna eq actually, thus i'm not able to actually realtime hear, what i want then. if i'm able to adust the numeric readout everything would be fine.

- finally a small, but very annoying bug/misbehaviour here:

when dragging the lowest band gain there sometimes occours a very loud "resonance" at the adjusted frequency.
it's much louder than the actual signal.
it sounds like a selfresonating band, near a distorted triangle wave. when adjusting and listening carefully concentrated to the adjustments this really totally makes me jump off my seat ... and it could easily damage the speakers in certain circumstances.

compressor:

i think this might also adress egberts problem ...
- an additional gain slider _before_ the saturation.
the reason is the same as stated on the eq issue.
that way one could control the input of the saturation.



other than these issues, again, thanks for thes plugs.
they could easily become a place on most of all the tracks in my songs. i simply love the possibility to peakstop after eq/compression without having to stick another plug after the eq/compressor.
while some good compressors allready are capable of that, there are just a few (not too good) eq's that can do that.
so thanks again!!
keep up!

[stefancrs]

Whoha! Great post brok!

hey stefan,
i've read the whole thread, and i've been trying out your plugins.
first, let me say that these are really nice!
thank you so much for releasing them for free!
may i still mention some things, that could help to get to the state these plugs deserve:

Why indeed, ofcourse you may, I encourage anyone to do so

eq:

- give the eq an input and a gain slider (the gainslider should be placed in frot of the saturation stage)
here is why i suggest the gain slider:
when eq-ing a signal, it can easily come to the situation that the level of the signal becomes loud, when boosting the bands. now, when i don't use the saturation, i don't have the possibility to readjust the output other than using the fader on the mixer. i don't like to use that mixerfader for this, as i need the full range to be able to ajust the signal precisely in the mix.
if i use the saturation, it could be that the signal gets way too hot, even if the saturation knob is set to zero, because of the signal being boosted way over the internal threshold of where the saturation starts to take place.
here is why i suggest the input slider:
some plugins are (like your's) not able to control the output, and sometimes they produce too high leveled output. placing your eq after it would set me to the need to lower the input before actually processing any further.

I _think_ you've made a mistake here. Isn't what you'd need for how you want it to work a chain like this:
input gain -> eq -> saturation -> output gain ?
You'd then use input gain to control the amount of saturation and the output gain for pure output volume control? If you'd put the output gain before the saturation the input and output gains would in effect do exactly the same thing to the signal, no?

- numeric readout-dragging/editing.
here's why i want that:
i (for one) like to be able to edit rather via numbers than dragging the graphics, as this is too unprecise for me.
also, when eq-ing the very high content of the sound it could happen, that the q-node runs out of the display. to edit the q then i have to move the frequency node to a deeper point in the frequency just to adjust it ... that at a frequency that i don't wanna eq actually, thus i'm not able to actually realtime hear, what i want then. if i'm able to adust the numeric readout everything would be fine.

Would non-editable readouts but "double" Q handles be sufficient? That is, you'd be able to edit the Q in both ends of that "Q-line". Would it be even better to have a key modifier instead, for instance so that you could hold ctrl when adjusting a band and then you'd adjust Q instead of gain/frequency?

- finally a small, but very annoying bug/misbehaviour here:

when dragging the lowest band gain there sometimes occours a very loud "resonance" at the adjusted frequency.
it's much louder than the actual signal.
it sounds like a selfresonating band, near a distorted triangle wave. when adjusting and listening carefully concentrated to the adjustments this really totally makes me jump off my seat ... and it could easily damage the speakers in certain circumstances.

What mode / Q / frequency / gain are you using when this is happening?

compressor:

i think this might also adress egberts problem ...
- an additional gain slider _before_ the saturation.
the reason is the same as stated on the eq issue.
that way one could control the input of the saturation.

The output gain _is_ before the saturation For me the saturation is mainly used to absolutely make sure no peaks / transients goes above 0dB.

other than these issues, again, thanks for thes plugs.
they could easily become a place on most of all the tracks in my songs. i simply love the possibility to peakstop after eq/compression without having to stick another plug after the eq/compressor.
while some good compressors allready are capable of that, there are just a few (not too good) eq's that can do that.
so thanks again!!
keep up!

Many thanks for the feedback! Without you and others like you it'd be much harder to tweak these towards completion!

[sinkmusic]

- The gui is good for both.
- the eq lacks some basic presets (and maybe so does the comp imo)
- one FR for the eq : a "disable" feature for each band is needed.

[bduffy]

Actually, I think the EQ has the presets from the compressor? "Limit 1", "Limit 2", etc.

[brok landers]

Whoha! Great post brok!

thank you, great plugins, thats why i posted ...

I _think_ you've made a mistake here. Isn't what you'd need for how you want it to work a chain like this:
input gain -> eq -> saturation -> output gain ?
You'd then use input gain to control the amount of saturation and the output gain for pure output volume control? If you'd put the output gain before the saturation the input and output gains would in effect do exactly the same thing to the signal, no?

yeah, you could also do it like this.
but my idea behind my post is simple:
every algorithm works better in calculation terms when the input is as high as possible.
the lower the signal is, the more calculation-errors take place, especially when there is no oversampling.
so here's my thinking:
1. input control is to adjust the incoming signal to the desired gain (ideally 0db), as the better the input level of the in-signal is adjusted, the more accurate the algorithm can work, correct?
ideally every signal should be as close as possible (not over, and not too deep under) 0db.
so that adjustment has nothing to do with the saturation at that stage (in my way of thinking).
2. then i eq the signal. lets assume (just theoretical), by equing i boost the signal to +4 db over zero.
now, if i'd send this direct into the active saturation stage, it would actually saturate the signal, even if it shouldn't, and even if the saturation slider is at minimum, because i've shot the signal over the internal threshold.
here is where i'd place the gain-slider.
i could then re-adjust the eq'ed signal (after being eq-ed at maximum level for max quality and most less calculation error) to 0db.
that allowes me to the use the saturation also at max resolution (0db), on top of that i can "dose" the saturation from absolutely subtle to max in the desired slider resolution.
after the saturation stage one would not need a gain slider, assuming the saturation won't generally not allow any overshoots when turned on.
and when the saturation is not used, the gainslider _before_ the saturation could keep track of the outputlevel, as it's the last stage of the chain then anyway ...

Would non-editable readouts but "double" Q handles be sufficient? That is, you'd be able to edit the Q in both ends of that "Q-line". Would it be even better to have a key modifier instead, for instance so that you could hold ctrl when adjusting a band and then you'd adjust Q instead of gain/frequency?

well, it would help of course, as i am then actually able to adjust the q the when frequency is set to max ...
but i am more into adjusting numerical (vial dragging up and down), it is simply more accurate than dragging graphically ...
i for one found it allways harder to achieve the desired gain and frequency by numrerical value changing, as to drag 2 dimensional in the graphics ... one time the gain is correct but not the gain, and while i try to get the gain ready, i accidentialy readjust the frequency and vice versa ...
to me the graphics are more to have a visual reference ...
ideally i find that if you implement both, every way to work will be satisfied.

What mode / Q / frequency / gain are you using when this is happening?

i cannot tell you as i'm at home right now ...
only 4 things:
1. it's only the lowest band
2. it's only on deep frequencies
3. it's allmost allways, but _only_ when actually dragging
4. it seems to be around +-0db, not on very much higher/lower gainsettings ...

The output gain _is_ before the saturation For me the saturation is mainly used to absolutely make sure no peaks / transients goes above 0dB.

ahh ... silly me ... srry for mentioning then ...

Many thanks for the feedback! Without you and others like you it'd be much harder to tweak these towards completion!

the pleasure is on my side!
btw, are you the developer of synthmaker?
or are you developing _on_ synthmaker?

[brok landers]

- one FR for the eq : a "disable" feature for each band is needed.

ahh ... yes ... i forgot to mention thisone ...
hehe ... we're getting hard on stefan now ...

[stefancrs]

- The gui is good for both.
- the eq lacks some basic presets (and maybe so does the comp imo)
- one FR for the eq : a "disable" feature for each band is needed.

Ok. Easily taken care of. Didn't know people ever used presets in an EQ, that's why I haven't made any Will add "disable" mode for the bands as well.

Actually, I think the EQ has the presets from the compressor? "Limit 1", "Limit 2", etc.

Just the names. The actual settings are probably bogus (have never changed preset via a host in the EQ )

[stefancrs]

I _think_ you've made a mistake here. Isn't what you'd need for how you want it to work a chain like this:
input gain -> eq -> saturation -> output gain ?
You'd then use input gain to control the amount of saturation and the output gain for pure output volume control? If you'd put the output gain before the saturation the input and output gains would in effect do exactly the same thing to the signal, no?

yeah, you could also do it like this.
but my idea behind my post is simple:
every algorithm works better in calculation terms when the input is as high as possible.
the lower the signal is, the more calculation-errors take place, especially when there is no oversampling.
so here's my thinking:
1. input control is to adjust the incoming signal to the desired gain (ideally 0db), as the better the input level of the in-signal is adjusted, the more accurate the algorithm can work, correct?
ideally every signal should be as close as possible (not over, and not too deep under) 0db.
so that adjustment has nothing to do with the saturation at that stage (in my way of thinking).
2. then i eq the signal. lets assume (just theoretical), by equing i boost the signal to +4 db over zero.
now, if i'd send this direct into the active saturation stage, it would actually saturate the signal, even if it shouldn't, and even if the saturation slider is at minimum, because i've shot the signal over the internal threshold.
here is where i'd place the gain-slider.
i could then re-adjust the eq'ed signal (after being eq-ed at maximum level for max quality and most less calculation error) to 0db.
that allowes me to the use the saturation also at max resolution (0db), on top of that i can "dose" the saturation from absolutely subtle to max in the desired slider resolution.
after the saturation stage one would not need a gain slider, assuming the saturation won't generally not allow any overshoots when turned on.
and when the saturation is not used, the gainslider _before_ the saturation could keep track of the outputlevel, as it's the last stage of the chain then anyway ...

Ok, uhm The eq algorithm doesn't perform better or worse even if the signal is very low/very high. That's one of the advantages of floating point maths. In a simplified manner one can say that the EQ is just as precise if the signal is in the -0.001 to 0.001 range as it does if the signal is in the -1000 to +1000 range.

When adjusting the "saturation" knob you actually sets the level at where the saturation begins to take place. When the knob is set at zero, the saturation starts coloring the sound when the signal gets close to +/- 10000000. Which it most likely does not. If the input is at 0dB and you boost it +20dB along the entire spectra you have a signal in the +/- 10 range. If you put all boosts at the same frequency you end up boosting that particular frequency 80dB, and have hence moved the signal into the +/- 1000 range. Still very far from where the saturation would kick in. But when not having the saturation at 100% you can get signals well above 0dB after the saturation stage, that's why I thought an output gain _after_ the saturation stage and one before (input gain) would be of interest. Otoh, the input gain won't really do much good either since you can still tweak the amount of saturation using the knob. The saturation knob really works as a saturation pre- AND post-gain. It pulls up the signal after the saturation stage just as much as it pulled it down before the stage, so you end up with a 0dB boost all-in-all, but with different amounts of saturation activity going on inside the stage

Would non-editable readouts but "double" Q handles be sufficient? That is, you'd be able to edit the Q in both ends of that "Q-line". Would it be even better to have a key modifier instead, for instance so that you could hold ctrl when adjusting a band and then you'd adjust Q instead of gain/frequency?

well, it would help of course, as i am then actually able to adjust the q the when frequency is set to max ...
but i am more into adjusting numerical (vial dragging up and down), it is simply more accurate than dragging graphically ...
i for one found it allways harder to achieve the desired gain and frequency by numrerical value changing, as to drag 2 dimensional in the graphics ... one time the gain is correct but not the gain, and while i try to get the gain ready, i accidentialy readjust the frequency and vice versa ...
to me the graphics are more to have a visual reference ...
ideally i find that if you implement both, every way to work will be satisfied.

Ok, point taken

What mode / Q / frequency / gain are you using when this is happening?

i cannot tell you as i'm at home right now ...
only 4 things:
1. it's only the lowest band
2. it's only on deep frequencies
3. it's allmost allways, but _only_ when actually dragging
4. it seems to be around +-0db, not on very much higher/lower gainsettings ...

Weird, will see if I can trigger the same behaviour here later on.

Many thanks for the feedback! Without you and others like you it'd be much harder to tweak these towards completion!

the pleasure is on my side!
btw, are you the developer of synthmaker?
or are you developing _on_ synthmaker?

It's all a bit blurry, innit? I AM on the SynthMaker team, so I am one of the developers, but I almost entirely focus on content (it's the dsp stuff I enjoy the most ) and such.

[stefancrs]

- one FR for the eq : a "disable" feature for each band is needed.

ahh ... yes ... i forgot to mention thisone ...
hehe ... we're getting hard on stefan now ...

It's all splendid I dare say! Easier to get these to be good plugins then!

[sinkmusic]

Didn't know people ever used presets in an EQ, that's why I haven't made any

Yes.
That's just i'm lazy. I don't use presets, but if I just want to cut a bass, instead of switching off 4 time each band, and set the lowest one to "locut", i'd better load the "bass cut" preset, and just have to set the good frequency and Q...
Most of the time, I use eq to cut one or two range on each track, to cut uneeded freq, and sometimes also boost one or two..

[stefancrs]

Didn't know people ever used presets in an EQ, that's why I haven't made any

Yes.
That's just i'm lazy. I don't use presets, but if I just want to cut a bass, instead of switching off 4 time each band, and set the lowest one to "locut", i'd better load the "bass cut" preset, and just have to set the good frequency and Q...
Most of the time, I use eq to cut one or two range on each track, to cut uneeded freq, and sometimes also boost one or two..

Yeah ok, I will add a few "starter template" presets when I get to it, I have a new release of the compressor coming out first, and it's a good'un if I may say so myself