I have been reading a lot about auto release, including a 66 page report on compression in general, and a lot of auto-releases seem to behave differently, and people say different things about them.
Some people say auto release is measuring the crest factor vs RMS, some people say it's basically just a curve on the release (as in the closer it gets to no gain reduction, the slower it gets), and U-He simply says that shorter percussive sounds lead to shorter release times.
I've been testing the auto-release a lot, and I have found no consistencies in any of these descriptions. Maybe that's because Presswerk's VU meter isn't reliably showing instantaneous changes, but I'm having a hard time understanding what their auto release does because of that. It doesn't seem to react differently to louder signals, and it doesn't seem to react differently based on the length of the signals.
Has anyone figured this out or can someone from U-He describe this in further detail than the manual did?
How Does Presswerk's Auto Release ACTUALLY Work (adapt knob)?
-
- KVRer
- Topic Starter
- 12 posts since 24 Aug, 2016
-
- KVRian
- 1115 posts since 2 Oct, 2001 from Berlin, Germany
The 'auto-release' (we're using the term adaptive, since it's more precise on what goes on inside) is a combination of several methods. First of all, it takes the peak value into account, and also the current RMS, which is pretty much like determining the crest factor. It then calculates further from there be feeding the detected portion of the signal (everything that's over the threshold) into severall parallel lowpass filters / integrators (8 of these, afair). A weighting function then picks the one(s) having the highest momentary value, and this feeds the actuall envelope stage. It's like a waterfall procedure where the signal has to pass multiple tests. As a result, short bursts/transients pass long integrators without impact, thus only resulting in a short overall release. Long bursts, or steady-state signals carry enough energy to fill up all integrators so that the final release approaches the maximum integration time of these filters.
Here's an audio file:
http://u-he.com/Sascha/_kvr/noiseburst% ... 20mono.mp3
Amplitude plot:
It's 30 secs of steady white noise (-48dBFS), and some noise bursts at -3dBFS (10, 20, 50, 100, 200, 500, 1000ms duration). Presswerk at -48dB threshold, ratio 20:1, attack minimum, release center, adaptive at max.
Check what happens to the release phase; it recovers gradually slower the longer the input signals stays over the threshold.
I hope this helps clearing that up a bit.
Here's an audio file:
http://u-he.com/Sascha/_kvr/noiseburst% ... 20mono.mp3
Amplitude plot:
It's 30 secs of steady white noise (-48dBFS), and some noise bursts at -3dBFS (10, 20, 50, 100, 200, 500, 1000ms duration). Presswerk at -48dB threshold, ratio 20:1, attack minimum, release center, adaptive at max.
Check what happens to the release phase; it recovers gradually slower the longer the input signals stays over the threshold.
I hope this helps clearing that up a bit.
Sascha Eversmeier
drummer of The Board
software dev in the studio-speaker biz | former plugin creator [u-he, samplitude & digitalfishphones]
drummer of The Board
software dev in the studio-speaker biz | former plugin creator [u-he, samplitude & digitalfishphones]
-
- KVRer
- Topic Starter
- 12 posts since 24 Aug, 2016
This is incredibly helpful, and also very reassuring, as I just found out my 1176 emulation doesn't have program dependent release, and I've been doing a lot of reading that says that is the original analogue characteristic (and a big part of the sound).
Just a few more questions.
1) I've been doing some more testing with white noise, and this is from the manual:
Higher Adapt values
mean a wider 'window' around the nominal release time. Short, percussive signals will lead to
shorter release times, while smooth passages will make recovery much slower.
It seems like this is implying that there is a width of available adaptation around the release you choose, but this only seems to be working one way. Am I doing it wrong? I have a very short transient and a very long increase in white noise volume, but changing the adapt knob doesn't make the transients release time shorter. It just makes the steady state one longer. I'm just wondering if this is correct, and adapt only defines how much MORE release time can be added on, and not less?
I am doing a very similar test to what you're doing. The white noise is a very very very short, sudden burst (as short as my DAW's midi will let me play a white noise synth, which is 2-3ms) and the solid signal is about a full second. The adapt works perfectly for the solid but the transient doesn't get shorter or longer with the adapt knob, making me think the adapt knob isn't a window around the nominal release so much as a range defined using the release as a minimum value.
2) From the manual:
"Of course manually finding the best settings is important, but we tried to make Presswerk sound
great with any settings by including algorithms that automatically ‘ride’ the audio and adjust various
parameters. We didn’t take the idea too far, though – you can still control the following..."
If you'll indulge me, what kind of settings are we talking about? Program dependent values like the attack? Changes to the sidechain detection? Anything specific to the feedback mode? I'm just wondering because I'd love to know how this compressor interacts with audio, and I think the stuff behind the scenes is why we choose the compressors we love.
Just a few more questions.
1) I've been doing some more testing with white noise, and this is from the manual:
Higher Adapt values
mean a wider 'window' around the nominal release time. Short, percussive signals will lead to
shorter release times, while smooth passages will make recovery much slower.
It seems like this is implying that there is a width of available adaptation around the release you choose, but this only seems to be working one way. Am I doing it wrong? I have a very short transient and a very long increase in white noise volume, but changing the adapt knob doesn't make the transients release time shorter. It just makes the steady state one longer. I'm just wondering if this is correct, and adapt only defines how much MORE release time can be added on, and not less?
I am doing a very similar test to what you're doing. The white noise is a very very very short, sudden burst (as short as my DAW's midi will let me play a white noise synth, which is 2-3ms) and the solid signal is about a full second. The adapt works perfectly for the solid but the transient doesn't get shorter or longer with the adapt knob, making me think the adapt knob isn't a window around the nominal release so much as a range defined using the release as a minimum value.
2) From the manual:
"Of course manually finding the best settings is important, but we tried to make Presswerk sound
great with any settings by including algorithms that automatically ‘ride’ the audio and adjust various
parameters. We didn’t take the idea too far, though – you can still control the following..."
If you'll indulge me, what kind of settings are we talking about? Program dependent values like the attack? Changes to the sidechain detection? Anything specific to the feedback mode? I'm just wondering because I'd love to know how this compressor interacts with audio, and I think the stuff behind the scenes is why we choose the compressors we love.
