So is this a bug, or something I’m getting ‘wrong’ with my audio/CoreAudio/JACK settings, etc.?
Here I’ve got a simple project open with the Ardour 8.6.7 demo/trial version. -No plugins are on anywhere. In this example a ‘tight’, fader-automation change is not fast enough to adequately capture a transient (bass drum hit). Although not shown in the video, this effect occurs with a less-steep automation ‘curve’ as well, indicating that perhaps all fader automation here is being delayed by something in the/my system:
Details:
x86_64 Demo Version of Ardour 8.6.7
Macbook Pro 2011, A1286, 2 GHz Core i7
OS X El Capitan 10.11.6
I also tested Ardour 6.9 (a full version), which had the same issue.
But the really old version of Ardour 2.8.16 did NOT exhibit this effect. (It responded quickly and accurately.)
Also, the problem gets worse as the Buffer Size is increased.
(-It is never resolved with a smaller buffer size, -just gets worse as you increase it.)
The video that you posted must be a test case where you isolated the behavior that’s concerning you. Is that right? I can see adjacent frames in the video where the fader jumps from -inf to 0.0, and when it hits 0.0 dB the playhead is indeed in the second “trough” of the waveform.
But can you put this in context? What does a “real” project look like? Could this be solved with a compressor, expander, duck, or gate plugin rather than a fader automation? After all, couldn’t one say the purpose of a compressor is to move the fader more quickly than a human (or maybe in this case Ardour) can?
I’m a non-expert hobbyist, so apologies if I’m missing something.
Since Ardour 4.0 the fader is dezippered (no zipper noise, click,or AM modulation due to instant gain changes). The gain stage follow roughly a 25Hz exponential ramp (same for mute).
No, you’re not missing anything. -Thanks for your comment!
In a ‘real life’ scenario, I might want to use some drum track or sound and fade it in so fast that it cuts-off any and all noise before whatever sound I’m aiming to have present in the mix. Up until now, I’ve relied heavily on fader automation in these circumstances.
I suppose I have to move forward with region gain changes instead, as Robin @x42 suggested, or other plugins if applicable, yes.
@x42 now you’ve gone and done it - turned on my engineer brain. Does this mean the time constant is 1/25 or 40 ms? If so the time to get to 99% of the gain/mute transient is 120 ms? That’s about 5 samples at 44.1 kS/s. What will the gain/mute transition look like if the automation is longer than that? Does the gain transition scale with audio sample rate in Ardour?
Also if the Y-axis is a log scale, shouldn’t the gain envelope appear linear?
Concerning region gain, I use it all the time and it’s a very important feature for me.
The waveform follows a non-linear log-meter, similar to the IEC meter deflection [1]. The range [-50 … -40] dB takes up as much vertical space as [-40 … -30] dB; and the range [-20 … top] spans 50% of the overall meter.
It is an non-linear exponential scale of the log-scaled (dB) [absolute] level (note that signal level crosses zero, so log() does not work as is).
Yes region-gain is likely the best option. It also stays with the region when you move it.
And note that most plugins also filter parameter values internally to prevent zipper noise.
–
The idea is to leave the fader for the slower artistic changes during mixing.
While not directly applicable here, the following short video is a gem (it’s for Mixbus, but equally applies to Ardour):
Hm. That is a nice concise tutorial. The new thing I learned was the use of the range tool to adjust the region gain curve. I was an avid fan of this in another DAW whose name I can’t remember just now …