Should I get 100% cancellation?

If I invert the Phase of 2 identical tracks and insert a plugin on 1, should I get complete cancellation?
I thought ADC took care of this?

Depends on the plugin but most will, if only applied to one of the identical tracks, null any cancellation. However the difference might be so small as to not matter in a musical context. Applications for things like games it might matter.

ADC took care of what? The processing done by the Plugin…?

If the plugin changes the audio at all you won’t get cancellation because one track is no longer the identical material. If the plugin did NOTHING to the audio at all but just introduced a delay, cubase’s plugin delay compensation should align and null them. But I’ve never tested this actually works.

If the plugin is doing anything different to the signal then the answer in no.

Also keep in mind that Cubase is not 100% sample accurate.
Start points are rounded.
Even two identical tracks do not always cancel.


Personally, I’ve found that two tracks that are just WAV files do cancel 100% when one is phase-reversed. I did this because I wanted to check for errors when I had my first CD pressed. I took the original file I sent them and the ripped test CD file and placed them on adjacent tracks, phase-reversed one of them, and got pure silence.

This may not mean Cubase is ALWAYS sample-accurate, but it worked perfectly for me.

I record in Mid/Side quite frequently… doing a session in about half an hour actually with a lovely Taylor acoustic and i usually check to see that the two side channels null… so far they ALWAYS have!
I use the ‘duplicate, pan hard L/R and phase invert one of the side channels’ method… works both before and after tracking… Obviously though adding a processor to just one of the side channels CAN cause issues as described by Lights and Split.

This would scare me a little. I’ve never experienced something like this. When does it happen?

Non-cancellation in general should not scare anyone making MUSIC. It’s a good test of calibration but normal / traditional music should work fine with it around. After all has anyone ever heard two drummers cancel each other out? Mind you I know I’m getting timing right when my playing cancels the metronome and I can no longer hear it.
Ask yourself. Is it essential? Even for dance, trance, industrial etc.
And in the real world I’ve only ever heard it used on test tones and never on stacked drums where the inaccuracies add to the character of the end result. In fact, on stacked drums I’d hate to think they were all at the exact same beat because that would mean they would be cancelling themselves at some point.

Cancellation is for testing the equipment chain and not the music. Music should not be used to test the equipment chain as there are more accurate test tones provided for that.

There’s a joke in there somewhere? :bulb:

I am surprised nobody has claimed they can hear the non-cancellation and that it is affecting their workflow and costing them clients. And that it is a bug and needs to be fixed ASAP, or else!

That’s bass players… :mrgreen:

Some have but really they have the wrong idea. On audio tracks it will always work. Has for me and matjones and lights so far.
Some complain about copied midi tracks. Synths and drum machines, even digital internal ones, will always have a degree of variation due to the midi packets thru the computer hitting various obstacles. Even when a midi track is copied to audio there will still be a variation in the original midi track if you try to cancel them against one another. And even if you copy the original twice!
I’ve yet to see a client who cares that much about cancellation at the recording stage. Studio engineers aren’t usually that bothered (although at top end I have seen a producer lock out the entire studio staff from a session because he really was bothered but I don’t think there’s many millionaires like him here) but the maintenance techs are as they set up the whole chain as close to optimum as possible.

It’s not just phase-cancellation you have to worry about. Getting that right is only some fraction of the studio build process. It’s pointless being bothered at all about it if you haven’t sonically measured your room and optimised your speaker placements (which need to be in phase too) and monitoring position etc.

Well, most of the time. it’s OK.
In earlier versions I did test it, by splitting the event on one of the tracks.
Not tested this for long.

It was confirmed by Steinberg.

Important :
The worst case offset is 1 sample !
That is really not affecting any musical timing relations !

there are cases, where it can be quite an issue :
Think of a drum overhead track ( or a mixed drum / drum loop ) where You make a copy
on a 2nd track, to do a different, additional processing to certain parts.
When You cut these parts, some may get that 1 sample offset.
This can cause noticeable high freq. cancellation …

when crossfading two events of the same audio file with high freq. content,
the fade area can sounded dull, due to the 1 sample offset.

Admittedly these are rare cases …
And I didn’t do any testing for some time.


LOL don’t start conman… i have a VAST catalogue of drummer jokes :wink: he he :laughing:

Heard them ALL. Drum solo over… bass player’s turn.

Drummer wakes in hospital, can’t remember anything. Doc says “Good news and bad” “Wassa bad news?” We had to take half your brain out." Wassa good news?"
“We bought you a bass guitar.” :mrgreen:

Why only a “Submit” button. Surely there should be a “Claim victory” button. :mrgreen:

Sample accuracy again: it’s surely not a musical problem. It can be essential for analyzing material and what ‘scares’ me here (the word is an exaggeration of course :sunglasses: ) is the fact that reliable results would not be possible, null tests were pointless.

To give you a real world example, remember that 3rd party plugins do not always work under certain circumstances after reopening a project. If it happens that a client wants something changed in a mix weeks or month after you actually finished his work, a null test from the actual mix vs. the bounced mix is helpful to reassure you’re going on from the latest situation and can easily find problems. If anything works I got either a blank null (which could be considered as a proof for sample accuracy) or unsynced modulations left over if in use (no proof possible).

Where did Steinberg confirm a non-accuracy issue?

So what about parallel mastering in Cubase?
if I process one track differently I am introducing
phase problems, I have tried sending this track to a buss but the sane think happens, is there
a plugin that can tell me how many samples my tracks has moved?

If you’re worried that the PDC isn’t compensating properly try switching the plugin bypass on for all FX in the chain (not off but bypass) this should switch the FX off but keep the plugin delay, then do a null test.