DynCurve – a different approach to dynamics processing

[jonasfehrenberg]

Hi all,

DynCurve is now officially released

Some of you might have seen it during the beta here on KVR — thanks again to everyone who tested and shared feedback.

DynCurve is a curve-based dynamics processor:
instead of threshold/ratio, you directly map input to output using a freely editable transfer curve.

This effectively unifies compression, expansion, upward/downward processing, and gating in a single system.

One thing that was important to me:
the display behind the curve shows the actual sidechain signal, so you’re always working on exactly what the processor “sees”. Attack, release, rms, hpf - behavior are reflected visually in that signal as well.

A few highlights:

* fully editable transfer curve
* precise control over specific level ranges
* dual-release system (for more complex release shapes)
* multi-channel support (incl. mid / side processing on multichannel busses)
* can be used both surgically and creatively

The goal was to move away from threshold/ratio/makeup gain - thinking and towards directly shaping dynamics.

Would love to hear what you think about the workflow and concept.

Download / info:
https://fehrenberg-audio.com/dyncurve.html

14-day trial, no account needed. Just download and activate.

[psy dive]

That does look tasty!

May I ask, why have you chosen to visualise the audio from top to bottom instead of sideways? Wouldn't sideways be much easier to see what part of the signal your expanding/compressing, given you'd scale the visual accordingly?

[billinder33]

May I ask, why have you chosen to visualise the audio from top to bottom instead of sideways? Wouldn't sideways be much easier to see what part of the signal your expanding/compressing, given you'd scale the visual accordingly?

I made that comment in the beta thread. Not to speak for the dev, but I'd guess the reason is to better correlate the metering on the curve as the audio passes by. Maintaining the synch of the curve feedback with the audio running horizontal would require rotating the curve graphic 90 degrees. But that would probably make even less sense and make the curve editing more difficult. The challenges of building metering against XY graphs.

But I agree, it's tough for me visualize. I feel the same way about the vertical metering in Smart:Limit and Krafter - very hard to wrap my mind around.

[jonasfehrenberg]

That does look tasty!

May I ask, why have you chosen to visualise the audio from top to bottom instead of sideways? Wouldn't sideways be much easier to see what part of the signal your expanding/compressing, given you'd scale the visual accordingly?

Thanks!
I decided to visualize the audio vertical because I wanted the whole Model to be logical in itself. It is a coordinate system with input on the x-axis and output on the y-axis. This is actually the way we already know how compressor-curves look like. I wanted to be consequent, so to visualize the input on the y-axis then does not make a lot of sense, although all the compressor plugins do so.
This way you actually see what the processor sees... and I promise you that it will make a lot of sense once you get used to it.

[kernaudioio]

the vertical sidechain display makes sense once you see it as a consequence of the coordinate system rather than a design choice - input on x, output on y, and you're literally watching the signal travel down the x-axis in real time. billinder's right that rotating it would break the curve editing.

the transfer curve approach is interesting from a control theory angle too. threshold/ratio is just a piecewise-linear approximation of an arbitrary transfer function with some convenience parameters bolted on. going straight to the curve removes the indirection. curious whether you ran into challenges keeping the curve smooth enough to avoid artifacts at inflection points, especially with fast attack settings.

very nice mate!

[jonasfehrenberg]

the vertical sidechain display makes sense once you see it as a consequence of the coordinate system rather than a design choice - input on x, output on y, and you're literally watching the signal travel down the x-axis in real time. billinder's right that rotating it would break the curve editing.

the transfer curve approach is interesting from a control theory angle too. threshold/ratio is just a piecewise-linear approximation of an arbitrary transfer function with some convenience parameters bolted on. going straight to the curve removes the indirection. curious whether you ran into challenges keeping the curve smooth enough to avoid artifacts at inflection points, especially with fast attack settings.

very nice mate!

Thanks for that educated answer. Actually it took a few tries, and as I am more musician and audio engineer that a mathematician, I had to do some research to find a method for interpolation without overshots at some edge cases. In the end it worked

[jonasfehrenberg]

Hi all,

quick update: DynCurve has been updated to version 1.0.2 with a few new features and workflow and usability improvements — many of them based on feedback from users and beta testers.

The bigger one first: Undo is now implemented.
I know this should be a basic feature, but since the curve editor is fully custom-built, this took a bit more work than a normal parameter undo, but curve edits and parameter changes can now be undone using the usual DAW shortcuts.

There is also a new Curve Ruler:
right-click inside the curve editor and DynCurve shows a ruler/readout for the exact input level, output level and gain difference at that point on the transfer curve. This should make it much easier to inspect what the curve is actually doing in specific level ranges.

Segment shape switching has also been improved.
Instead of using right-click for spline/linear switching, small inline segment controls now appear when hovering near a curve segment. So switching between smooth and linear segments is much more direct now.

The Output section has a new Level Match feature.
DynCurve continuously compares input and output and shows the difference as Delta True Peak and Delta RMS. Each has its own Match button, so you can instantly adjust the Output Gain to match either peak level or RMS energy at the moment you are listening to.
This is meant to complement Auto Gain: Auto Gain gives a quick estimate based on the curve, while Level Match uses the actual measured signal.

Also new:
- click-free bypass for smooth A/B comparisons
- improved UI rendering, especially with larger buffer sizes
- smoother curve/envelope display without the previous stepped movement

Download / info:
https://fehrenberg-audio.com/dyncurve.html

As always, there is a 14-day trial, no account needed.

Thanks again for all the feedback — a lot of this update came directly from things people pointed out while using the plugin.