2

Phase cancelation between different audio tracks is an issue that cannot be solved just by a constant phase offset. There are sweet spots, but no matter what the global relative phase of two tracks is, there will always be some frequencies interfering destructively, while others might become too dominant due to constructive interference. I've found that this is the major factor, why two instruments mask each other when combined. As the phase relations change over time, there is also an inconsistency in the timbre, which I don't like.

So I wonder, if there are strategies to somehow even out the intereference effect. Is there a plugin to manipulate phase as a function of frequency... just like an EQ, but for the phase instead of the amplitude. Is there a plugin to define a desired phase relation and have the phases a different tracks aligned automatically for each frequency band? That would give interesting options: A phase offset of 0° would make two signals fundamentally additive. A phase offset of 90° would highlight the complementary parts of two signals by avoiding a buildup in the regions where they overlap. A phase offset of 0° would subtract the overlap of two signals, which is generally not a good thing, but might interesting as an effect.

Do you think that would work, or do you think the phase distortion would alter the timbre of each track so much that it would sound unnatural?

11
  • Do you mean the same audio recorded from two mics, stereo pair, or drum spill etc, or some kind of sample-based setup?
    – Tetsujin
    May 23 at 16:11
  • @Tetsujin no I mean different parts.
    – Fid Rewe
    May 23 at 16:34
  • 1
    How do you get phase cancellation between different parts? There should be almost zero correlation.
    – Tetsujin
    May 23 at 16:36
  • 1
    So you're suggesting the chance of a singer hitting a target note spot on is astronomically low? Come on, I can do it with my voice. Depending on the relative offsets of two good takes of the same note, I can boost or attenuate the first harmonic by a huge amount. I even can see, where I am in the waveforms.
    – Fid Rewe
    May 23 at 16:54
  • 1
    There are phase-correlation plugins. It would be worth running the tracks in question, soloed, centre-panned, through one strapped across the stereo output bus, to see if your perception is correct. I very much doubt it is. You could confirm by doubling one of the tracks & repeating the experiment with differing time-shifts, or even a modulated micro-delay.
    – Tetsujin
    May 23 at 17:38
2

Phase cancellation is really only an issue when there are several instances of the SAME instrument, playing the same notes. And, of course, the ultimate manifestation of this is multiple instances of the same sampled instrument. It's why we don't build up an orchestral violin section with 8 tracks playing the same notes, and why a good sample set has separate sample banks for Violin 1 and Violin 2. And why Automated Double Tracking (ADT) is of only limited use.

Are you having practical problems with mixing DIFFERENT instruments? Live, or sampled? The usual techniques are applying a small delay to one of them (often inaccurately referred to as 'shifting the phase'). Or some stereo separation can help.

Or maybe you're layering up bass instruments in a quest for the ultimate 'phat' sound? Particularly if they're all derived from similar, relatively simple sinewave-based methods of synthesis. This can cause problems. Sometimes less is more!

2
  • You might be right. I've noticed the problem when putting different vocal takes of the same passage on top of each other in order to make them sound more dense (artificial doubling sounds even worse to me). Is that something that doesn't work in general? However, it's not clear to me, how the effect is not supposed to happen to some extent among different instruments as well, as notes played together in song usually share a lot of overtones.
    – Fid Rewe
    May 23 at 16:47
  • It can be amazing how accurately a singer can imitate themselves in subsequent takes! Don't worry about whether it SHOULD theoretically happen between different instruments. Solve the problem if it occurs. Has it? May 23 at 17:44

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.