Any Phase Cancellation optimizer RE or VST?

Have an urge to learn, or a calling to teach? Want to share some useful Youtube videos? Do it here!
Post Reply
RobC
Posts: 1832
Joined: 10 Mar 2018

12 Sep 2019

When you mix two sounds together, it's more than likely that there will be at least some cancellation happening.

Has anyone ever seen such device, that compares two samples, and automatically offsets them, so there will be the least amount of cancellation?

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

12 Sep 2019

RobC wrote:
12 Sep 2019
When you mix two sounds together, it's more than likely that there will be at least some cancellation happening.

Has anyone ever seen such device, that compares two samples, and automatically offsets them, so there will be the least amount of cancellation?
There are three things to potentially address: timing, polarity, and phase. Timing involves things like latency, and there is ADC to address that in most cases. Polarity is pretty simple, and given there are two choices and in most cases there's no "Right" answer (except for with multi microphone setups), you're better off choosing for yourself. The last choice is phase, which many folks don't fully understand often thinking it's the same thing as polarity (it's not).

There are devices like Little Labs IBP that can address the phase issues, while Norman Henson's VMG-01 can address the timing issues, and you can use you're ears for polarity.
:)
Selig Audio, LLC

PeterP
Posts: 84
Joined: 26 Apr 2016
Location: Gothenburg, Sweden

12 Sep 2019

I use Harrison MixBus for this.

http://www.harrisonconsoles.com/mixbus/ ... -maximizer

If I'm mixing a rock song or something that has a multi miced acoustic drum kit then I just drag in the raw tracks into Harrison, select a section of the song where all the drums are hit, run the polarity optimizer, write down the results and manually apply it in the Reason mixer.

RobC
Posts: 1832
Joined: 10 Mar 2018

13 Sep 2019

selig wrote:
12 Sep 2019
RobC wrote:
12 Sep 2019
When you mix two sounds together, it's more than likely that there will be at least some cancellation happening.

Has anyone ever seen such device, that compares two samples, and automatically offsets them, so there will be the least amount of cancellation?
There are three things to potentially address: timing, polarity, and phase. Timing involves things like latency, and there is ADC to address that in most cases. Polarity is pretty simple, and given there are two choices and in most cases there's no "Right" answer (except for with multi microphone setups), you're better off choosing for yourself. The last choice is phase, which many folks don't fully understand often thinking it's the same thing as polarity (it's not).

There are devices like Little Labs IBP that can address the phase issues, while Norman Henson's VMG-01 can address the timing issues, and you can use you're ears for polarity.
:)
PeterP wrote:
12 Sep 2019
I use Harrison MixBus for this.

http://www.harrisonconsoles.com/mixbus/ ... -maximizer

If I'm mixing a rock song or something that has a multi miced acoustic drum kit then I just drag in the raw tracks into Harrison, select a section of the song where all the drums are hit, run the polarity optimizer, write down the results and manually apply it in the Reason mixer.
Thank both of you for the help. If it's okay, I reply in one comment:

Hmm, these rather seem to solve issues where the two or more samples are rather identical. What I meant is rather when specific frequencies of (for example) a kick, bass and snare interfere with each-other. (And let's rather think of simple 1 shot samples, that are much more easy to manipulate.)
With a basic solution, a tool would analyze the waveforms, and do a comparison, watching how much cancellation happens at what position, if the two are mixed together, with one being delayed (it would "sweep" through, side-by-side).

Of course, one problem is, that this can create a flam/attack effect ~ which, if desired, one should set themselves. The other problem is, that even if the tool finds the perfect setting, some desired frequencies still might drop out. Doing a Linear Phase multi-band split not only has artifacts, but even if offsetting multiple frequency bands might technically help, once the bands are rejoined, there's a good chance, the original sound will be pretty much musically wrecked.

So, there can be some heavy trade-offs with what I want. A side solution (also imperfect) might also be that two samples are again split into multiple frequency bands, a tool analyzes their loudness, calculates how loud each band ideally should be when mixed together, and once the two actually are mixed together, it would automatically compensate the loss, by adding either a static, or expander-style gain.

I've also looked at phase rotators, and while they can create headroom, and pretty much morph the given sample, it seems they create too much audible artifacts.

User avatar
NekujaK
Posts: 631
Joined: 09 Oct 2016
Location: USA

13 Sep 2019

I've used MAutoAlign from Melda Productions on multi-mic drum kits with great results.

https://www.meldaproduction.com/MAutoAlign
wreaking havoc with :reason: since 2.5
:arrow: https://soundcloud.com/nekujak-donnay/sets

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

13 Sep 2019

RobC wrote:
13 Sep 2019

Thank both of you for the help. If it's okay, I reply in one comment:

Hmm, these rather seem to solve issues where the two or more samples are rather identical. What I meant is rather when specific frequencies of (for example) a kick, bass and snare interfere with each-other. (And let's rather think of simple 1 shot samples, that are much more easy to manipulate.)
With a basic solution, a tool would analyze the waveforms, and do a comparison, watching how much cancellation happens at what position, if the two are mixed together, with one being delayed (it would "sweep" through, side-by-side).

Of course, one problem is, that this can create a flam/attack effect ~ which, if desired, one should set themselves. The other problem is, that even if the tool finds the perfect setting, some desired frequencies still might drop out. Doing a Linear Phase multi-band split not only has artifacts, but even if offsetting multiple frequency bands might technically help, once the bands are rejoined, there's a good chance, the original sound will be pretty much musically wrecked.

So, there can be some heavy trade-offs with what I want. A side solution (also imperfect) might also be that two samples are again split into multiple frequency bands, a tool analyzes their loudness, calculates how loud each band ideally should be when mixed together, and once the two actually are mixed together, it would automatically compensate the loss, by adding either a static, or expander-style gain.

I've also looked at phase rotators, and while they can create headroom, and pretty much morph the given sample, it seems they create too much audible artifacts.
There really isn't a solution because there isn't a "problem" IMO - random phase is what makes music sound like music. If you could eliminate it, I would guess things would sound pretty horrible.

For one thing, it's constantly changing. Even with identical audio clips in the timeline, the phase relationships evolve, and when you "fix" one you cause another. There isn't one perfect setting that would solve every phase variation - complex waveforms cannot be "conformed" in this way since they are bi-polar - always going in/out of phase at audio rate!

The only situation where you can absolutely correct for phase is with two of the same exact waveforms - frequency cannot drift, level cannot change, timing cannot change even one sample. And those issues can be compensated with by using delay compensation. All other issues are a tossup IMO.
Selig Audio, LLC

RobC
Posts: 1832
Joined: 10 Mar 2018

14 Sep 2019

NekujaK wrote:
13 Sep 2019
I've used MAutoAlign from Melda Productions on multi-mic drum kits with great results.

https://www.meldaproduction.com/MAutoAlign
Neat, looks like it does exactly what I've imagined.
Meanwhile it seems, what I would have used such device for, wouldn't work the way I imagined.
selig wrote:
13 Sep 2019

There really isn't a solution because there isn't a "problem" IMO - random phase is what makes music sound like music. If you could eliminate it, I would guess things would sound pretty horrible.

For one thing, it's constantly changing. Even with identical audio clips in the timeline, the phase relationships evolve, and when you "fix" one you cause another. There isn't one perfect setting that would solve every phase variation - complex waveforms cannot be "conformed" in this way since they are bi-polar - always going in/out of phase at audio rate!

The only situation where you can absolutely correct for phase is with two of the same exact waveforms - frequency cannot drift, level cannot change, timing cannot change even one sample. And those issues can be compensated with by using delay compensation. All other issues are a tossup IMO.
I thought a lot about this, and read an article on Soundonsound (about mixing bass frequencies, in short, it rather suggested watching the arrangement, trying to avoid frequencies overlap too much - so the common practice).

For me, the main thing that I didn't like in music was, when quite often a kick, or snare hit, that's supposed to pop out, kind of sounds flat or drowns in the mix. And this happens in any kind of music. Now, it can be expanded manually even with a simple mix fader automation (adding micro dynamics - which is a hassle, but I find it sounds pretty cool in the end) - but I kind of had a bad feeling there when doing that, since all the sounds mixed together obviously have a very varied phase in varied frequencies. Then again, it's not like during mastering, the final mix' waveform's volume wouldn't get touched, when compressing processing is done. Still, I did wonder if there's a cleaner / balanced approach to this - but at the end of the day I realized, all the processing I thought of, would actually just add more dirt.

All in all, thank you for clearing that up!

boomer
Posts: 75
Joined: 09 Nov 2016

14 Sep 2019

I think you are assuming that phase shift is constant, but it almost never is. So the question becomes match phase to what?

RobC
Posts: 1832
Joined: 10 Mar 2018

15 Sep 2019

boomer wrote:
14 Sep 2019
I think you are assuming that phase shift is constant, but it almost never is. So the question becomes match phase to what?
The thought came from a video actually, where a guy tried various ways to make a kick and a snare work together with as little frequency dropouts as possible, but there's only so much one can do, and there's a trade off with everything - which isn't worth the effort.

boomer
Posts: 75
Joined: 09 Nov 2016

15 Sep 2019

Well you can do that to a point. But that’s probably done best by ear because it involves your definition of best tone. And that may not be the same as maximum level.

And you may not know this, but every time you adjust EQ you are shifting phase somewhere across the frequency band. So phase shift itself isn’t necessarily a problem but instead just one of many factors.

User avatar
bik44
Posts: 98
Joined: 03 Jun 2015
Location: Poland

16 Sep 2019


User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

16 Sep 2019

bik44 wrote:
16 Sep 2019
I use Auto Align plugin
https://www.soundradix.com/products/auto-align/
Keep in mind these tools are designed for correcting delay on a multi microphone setup, and work because the same audio is basically in all microphones.
It would not work, for example, on a kick sample and bass synth playing at the same time (or similar) because there is no common reference to compare.
Selig Audio, LLC

RobC
Posts: 1832
Joined: 10 Mar 2018

16 Sep 2019

boomer wrote:
15 Sep 2019
Well you can do that to a point. But that’s probably done best by ear because it involves your definition of best tone. And that may not be the same as maximum level.

And you may not know this, but every time you adjust EQ you are shifting phase somewhere across the frequency band. So phase shift itself isn’t necessarily a problem but instead just one of many factors.
True, and I usually like to do a little offsetting either with regroove or delaying, to get a flavor of 'attack' or 'flam', and that always goes by the rule of whatever sounds best anyway.

Usually what I do: Kick stays, bass goes slightly after kick, snare goes behind kick, hat goes behind snare, polyphonic lead gets randomized timing with regroove, until it doesn't sound off ~ and similar with evolving pads; etc.

An interesting thing is that in that video I remember the guy trying high pass filtering on already-filtered-low-frequencies to shift the phase.

However, I rather use Linear Phase EQs, which may not be perfect either, not even when it comes to phase, but they do the trick. Of course, like I said, I don't want to do any changes like that either.

User avatar
TritoneAddiction
Competition Winner
Posts: 4219
Joined: 29 Aug 2015
Location: Sweden

16 Sep 2019

RobC wrote:
14 Sep 2019
I thought a lot about this, and read an article on Soundonsound (about mixing bass frequencies, in short, it rather suggested watching the arrangement, trying to avoid frequencies overlap too much - so the common practice).

For me, the main thing that I didn't like in music was, when quite often a kick, or snare hit, that's supposed to pop out, kind of sounds flat or drowns in the mix.
I know it's not what you originally asked about, but to me it seems like Carve EQ Ducker and Disperser could help you out.

Carve is great for avoiding too much frequenciy overlap between two sounds. I use it all the time.
And Disperser is great for making drums pop out more in the mix, without having to raise the volume. It changes the phase of the transients somehow. It doesn't always work, but sometimes it does.

RobC
Posts: 1832
Joined: 10 Mar 2018

17 Sep 2019

TritoneAddiction wrote:
16 Sep 2019

I know it's not what you originally asked about, but to me it seems like Carve EQ Ducker and Disperser could help you out.

Carve is great for avoiding too much frequenciy overlap between two sounds. I use it all the time.
And Disperser is great for making drums pop out more in the mix, without having to raise the volume. It changes the phase of the transients somehow. It doesn't always work, but sometimes it does.
I used to do something like Carve with a tool I built, although I gotta say, recently I like to minimize processing, and go the more natural way. It does make me think again about it, though.
I read about Disperser, though didn't really check that one out yet.

User avatar
TritoneAddiction
Competition Winner
Posts: 4219
Joined: 29 Aug 2015
Location: Sweden

17 Sep 2019

RobC wrote:
17 Sep 2019
TritoneAddiction wrote:
16 Sep 2019

I know it's not what you originally asked about, but to me it seems like Carve EQ Ducker and Disperser could help you out.

Carve is great for avoiding too much frequenciy overlap between two sounds. I use it all the time.
And Disperser is great for making drums pop out more in the mix, without having to raise the volume. It changes the phase of the transients somehow. It doesn't always work, but sometimes it does.
I used to do something like Carve with a tool I built, although I gotta say, recently I like to minimize processing, and go the more natural way. It does make me think again about it, though.
I read about Disperser, though didn't really check that one out yet.
Well try them out if you want. All I know is that these two tools can't be replaced with anything else (at least that I know of).
The thing with both of these REs is that it's easy to overdo it. A little goes a long way.

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

17 Sep 2019

RobC wrote:
16 Sep 2019
Usually what I do: Kick stays, bass goes slightly after kick, snare goes behind kick, hat goes behind snare, polyphonic lead gets randomized timing with regroove, until it doesn't sound off ~ and similar with evolving pads; etc.
Just curious what your goal is in doing the above?
I've never gone that deep to make a mix sound good, and have never felt the need to do so. But I also never stop learning and trying new things, so I'm curious what results you get from this approach. Could be I'm already addressing the issue another way, and so this wouldn't apply to my approach - but also could be something worth investigating… :)
Selig Audio, LLC

RobC
Posts: 1832
Joined: 10 Mar 2018

17 Sep 2019

selig wrote:
17 Sep 2019
RobC wrote:
16 Sep 2019
Usually what I do: Kick stays, bass goes slightly after kick, snare goes behind kick, hat goes behind snare, polyphonic lead gets randomized timing with regroove, until it doesn't sound off ~ and similar with evolving pads; etc.
Just curious what your goal is in doing the above?
I've never gone that deep to make a mix sound good, and have never felt the need to do so. But I also never stop learning and trying new things, so I'm curious what results you get from this approach. Could be I'm already addressing the issue another way, and so this wouldn't apply to my approach - but also could be something worth investigating… :)
In case of synthesized and programmed music, the perfectly timed samples can sound really weak when they hit together. Randomizing the timing of everything with ReGroove can make it more lively, but that can result in, that during the arrangement, some simultaneous sample hits sound great, others not so much. I prefer not to leave that to luck, but instead set it manually the way I said. Then I have full control over how the attack/flam effect they create, sounds. Yes, that will still sound programmed, but I prefer that tuned timing over the randomized groove. In fact, I even start doubting that in case of polyphonic instruments, it would be a good idea to leave it to randomness.

Of course, it's not a must to time sounds exactly in the order I said in my previous comment, it's just what I was mostly satisfied with.

Another thing is, that even when an attack/flam effect doesn't really happen, it still can help to separate sounds a bit with timing ~ kind of like pre-delay in case of reverbs. Might seem subtle, but it can make quite a difference.

If changing timing doesn't do anything ~ say, a soft pad, or simple wind blowing effect, then it clearly can be left as is.

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

18 Sep 2019

RobC wrote:
17 Sep 2019

In case of synthesized and programmed music, the perfectly timed samples can sound really weak when they hit together. Randomizing the timing of everything with ReGroove can make it more lively, but that can result in, that during the arrangement, some simultaneous sample hits sound great, others not so much. I prefer not to leave that to luck, but instead set it manually the way I said. Then I have full control over how the attack/flam effect they create, sounds. Yes, that will still sound programmed, but I prefer that tuned timing over the randomized groove. In fact, I even start doubting that in case of polyphonic instruments, it would be a good idea to leave it to randomness.

Of course, it's not a must to time sounds exactly in the order I said in my previous comment, it's just what I was mostly satisfied with.

Another thing is, that even when an attack/flam effect doesn't really happen, it still can help to separate sounds a bit with timing ~ kind of like pre-delay in case of reverbs. Might seem subtle, but it can make quite a difference.

If changing timing doesn't do anything ~ say, a soft pad, or simple wind blowing effect, then it clearly can be left as is.
I get it, interesting approach!

I can confirm that, at least in my opinion, bass lagging slightly behind kick sounds great. it's a sound you've heard before, I would have previously called it "tight" until I started working on Pro Tools in Nashville, with some stupidly great studio musicians. I kept looking at the tracks that felt a certain way (that I liked), and over the decades I saw the same thing - kick on the beat and bass a few ms (5-15 ms no problem) sounded "tight".

The snare behind the kick is trickier, and I feel less important because kick and snare already sit with no overlap on the spectrum for the most part.
Selig Audio, LLC

boomer
Posts: 75
Joined: 09 Nov 2016

18 Sep 2019

Just sharing my experience as it is a bit different. What I found was way back in the early MIDI (24 ticks) days was by moving the kick and bass off the exact timing both instruments had more low end which I attribute to some phase cancellations.

However for me it worked best to put the bass directly on the beat and slip the kick 2 ticks behind. Usually the snare was set 4 ticks behind the actual beat. Of course it depends what kind of music you are doing, but this works great if your target is “in the pocket” say like the drums of Al Jackson Jr on “Green Onions”. Probably not so much for Euro Synth Pop :j

User avatar
selig
RE Developer
Posts: 11685
Joined: 15 Jan 2015
Location: The NorthWoods, CT, USA

18 Sep 2019

boomer wrote:
18 Sep 2019
Just sharing my experience as it is a bit different. What I found was way back in the early MIDI (24 ticks) days was by moving the kick and bass off the exact timing both instruments had more low end which I attribute to some phase cancellations.

However for me it worked best to put the bass directly on the beat and slip the kick 2 ticks behind. Usually the snare was set 4 ticks behind the actual beat. Of course it depends what kind of music you are doing, but this works great if your target is “in the pocket” say like the drums of Al Jackson Jr on “Green Onions”. Probably not so much for Euro Synth Pop :j
Interesting, that's exactly the opposite of what I observed when looking at the old school session musicians in Pro Tools. The bass was always behind for "in the pocket" feel. For one thing, the kick is a shorter sound with lots of attack, and the bass is typically longer with less transient attack (with some funk bass being the possible exception). I've always assumed if the kick hits first it's in and out, then the bass "takes over". This way they never actually overlap, and the kick becomes the attack transient for the bass creating one big sound. That's a simple way to have both bass and kick full volume without conflicting and thus needing ducking or other "tricks".
The other way around doesn't make any sense in this regard, or am I tripping?
Selig Audio, LLC

RobC
Posts: 1832
Joined: 10 Mar 2018

20 Sep 2019

selig wrote:
18 Sep 2019

I get it, interesting approach!

I can confirm that, at least in my opinion, bass lagging slightly behind kick sounds great. it's a sound you've heard before, I would have previously called it "tight" until I started working on Pro Tools in Nashville, with some stupidly great studio musicians. I kept looking at the tracks that felt a certain way (that I liked), and over the decades I saw the same thing - kick on the beat and bass a few ms (5-15 ms no problem) sounded "tight".

The snare behind the kick is trickier, and I feel less important because kick and snare already sit with no overlap on the spectrum for the most part.
I kind of had the feeling, that as usual, I just 'reinvent the wheel'. x D

Sometimes a kick and a snare can flatten out each-others transients/clicks (easily happens with synth samples), and while a polarity flip can solve that, I still like to check a bit of offsetting to hear if I can create a nice sounding attack/flam effect.
In other words, I mostly use this as a bit of a bonus sound design / arrangement trick.

In case of kick and bass, especially with synth sounds - looping solo hits of both, then together, can help finding the offsetting sweet spot.

Meanwhile I thought about polyphonic lead synths. Honestly, if it's 1 sample based, then there's a chance that they are so well in sync, that a two-fisted chord can sound better (or at least more interesting) perfectly in sync, than if they are offset. Of course, that doesn't really work with multi-sampled, or probably not even time-stretched synths.
boomer wrote:
18 Sep 2019
Just sharing my experience as it is a bit different. What I found was way back in the early MIDI (24 ticks) days was by moving the kick and bass off the exact timing both instruments had more low end which I attribute to some phase cancellations.

However for me it worked best to put the bass directly on the beat and slip the kick 2 ticks behind. Usually the snare was set 4 ticks behind the actual beat. Of course it depends what kind of music you are doing, but this works great if your target is “in the pocket” say like the drums of Al Jackson Jr on “Green Onions”. Probably not so much for Euro Synth Pop :j
I think we're all talking about the same thing, just word it differently. : ) Well, except that in your case the bass is on the beat ~ but the kick is offset the same way. Aka, the kick is heard first, then the bass (of course, we're talking milliseconds here).

Post Reply
  • Information
  • Who is online

    Users browsing this forum: No registered users and 8 guests