Offline vs Realtime Mix Sound Quality: A/B TEST

Cubase
1

Hello folks!

Discovered accidentally while testing other things, I have found that a direct A/B test of a section of audio going through a lot of mastering inserts sounds better with a Realtime bounce than Offline.

I don’t know if this is a general thing or perhaps a problem with some of the mastering inserts I’m using (Nebula, various Slate Digital plugins, Ozone 6). I would describe the sound difference as a slightly noticeable loss of detail. Anyone else noticed this?

A/B TEST. Which Sounds Better To You? One is real time, one is offline.

CLIP A:

CLIP B:

2
  1. Print both methods.
  2. Import both into a empty session.
  3. Make sure they are sample accurate aligned in time.
  4. Invert Polarity on one of them. SB (incorrectly) calls this “Phase” and it’s in the “Pre” section of the Mixer.
  5. If they cancel, there is no difference. If they do not, whatever you hear is the difference between them.

“Cancel” is a relative term, as it’s commonly used. By that, I mean the depth of cancellation. Two files can “cancel” only partially, but the difference can fall below the level of detection at normal monitoring level so that the listener calls it “cancelled.” True cancel is total absence of signal. You should use “Statistics” to check.

If there is a difference, it will almost certainly be the attack and release times of any compression you are using. You will hear spitty garbage mainly on attacks of drums, etc.That is one parameter that can definitely vary between Realtime and Offline, depending on the plugins. Universal Audio plugin for sure. Though I still love and use them.

3

They don’t cancel, but someone elsewhere has suggested that the difference might just be the ‘randomness’ introduced by ‘analog’ artefacts of the Slate plugins (VMR, VTM, VBC).

4

Any compression, saturation or otherwise dynamic plugin will always result in a slightly different render.

5

Some plugins have a “live” mode for realtime rendering that limits the CPU usage.
Some plugins use x amount of over sampling when doing offline bounces.
That usually means that realtime bounce can potentially sound worse, as it is limited to a fixed CPU use, and offline bounce has the benefit of Time x CPU use. A one minute song could render in 5 minutes and use better algorithms.
There absolutely can be a difference in sound, I personally prefer that my mix sounds exactly the same when rendered.

6

I know that with many Nebula libraries, you’ll get better results using real time renders.

7

You probably already know this, but any randomized FX like reverb, chorus and most delays will not cancel. You have to kill all of that in your mix first. Forget to mention that.