Volume automation sounds different on each playback.

Hi guys and gals!
I usually work with some really tricky vocals, and I find myself need to write some very fast and deep volume automations for the vocals.
What I noticed is, sometimes, the automations plays like it should, but sometimes they’re not reacting “correctly”, like on some fast sibilance ducks, sounds like it missed that spot.
Is the automation in Cubase 7.5 not reacting fast enough during real-time playback?

Is Click enabled for the output bus in VST Connections - Outputs? Just click on the Click.

Sorry - this reply is for another post…

Is this misposted in the wrong thread?^^^

OP - that happens to me too with automation that is too abrupt.

Some things that may be helpful:

  1. Split the offending section out onto its own track, then you can make the leading automation slower without affecting the preceding audio.

  2. Render the automation, non-real time. Can always render the automation in several steps/passes (have to be super attentive that all the automation was rendered completely, or you’ll be saying “What!!???” for for a long time when you play it back). Adjust buffer as needed, if making the buffer bigger doesn’t help, maybe making it a bit smaller might (I’ve read that can be helpful in this case, I’ve never tried that and can’t confirm).

  3. This one is just a suggestion I haven’t tried - render or freeze everything else so more of the computer power can be used for automation.

Yeah - sorry. I clicked the wrong post to reply to…

I don’t use Volume Automation for stuff like sibilance ducking. Instead I manipulate the level on the actual Audio Part. There’s 2 similar ways to do this. One is to cut the part just before (or after) the sibilance and drag the part’s fade in/out to tame the audio. This works best if there is some empty space, even if it is very short, before or after the offending audio since the fade will start or end at zero. I have a couple of custom fade curves to facilitate doing this. Even without some space this technique often works fine. But if the fade does create an unwanted artifact there is an alternative. Cut the Audio Part immediately before and after the sibilance and drag the top of this section down to lower the volume for just the duration of the problem.

The befit of doing it this way is it allows the Volume Automation to be focused on setting mix levels independent of trying to surgically adjust for sibilance, plosives and the occasional note sung too soft or loud. After a bit of practice it’s pretty quick to do since mostly you can visually tell where to cut and it is pretty forgiving about not getting the cut location exactly right.

Thanks for all the answers!

This is very annoying since it’s unpredictable. And it costs at least half of my time just going back and forth to make sure.
I always find myself automating the vocal fader after compressions and EQs, because the processing on the vocal and the instruments change the perceived volume of the vocal.

Here’s what I usually do on vocals if I want a smooth polished sound, tell me if I’m wrong (Most of the vocals I work on are VERY dynamic):

  1. Clean up: strip silence, cut and drag down some really loud words.
  2. Pocketing: check grooving, make changes if needed.
  3. Light compression: if the vocal is recorded without any tracking compressor, I’ll add a gentle low-cut, then a “tracking style” compressor for 3dB gain reduction, and bounce it to an audio track.
  4. Rough automation: make the volume changes more reasonable, and easier for the compressor.
  5. Busing: output the track to a bus.
  6. EQs: on the bus, start with low-cut, avoid all the fundamental frequencies. Tame some boxiness. Boost highs above 12k or so, add/dip the low end and mid range depending on the recordings.
  7. Compression: first a limiter just barely touching those rare loud peaks, then a compressor, opto type for slower tempo, VCA type for up tempo. 4:1 ratio mostly, somewhat fast attack, somewhat fast release if “in your face” effect is needed. 6dB gain reduction at the loudest part. Enable low-end sidechain if needed.
  8. De-essing: one or several de-esser are used to tame sibilance, at different frequencies if needed.
  9. Touch-ups: maybe another EQ, maybe another compressor even. Limiter if needed.
  10. Final automation: draw automations word by word, to make the vocal sit right, not too loud for the verse, not buried in the chorus.

Here’s an example for a very tamed vocal (lots of automation involved):

BTW, Cubase’s buses are great, I don’t need to worry about which bus did I send it to, and which bus it is listening to. Without losing complex routing options.