Phase Inversion with Club Music (Humor me with the idea, please)

[rb1138]

Hello everyone,

First-time topic! Pretty new member.

Not sure if this is the right forum to post this in, but I thought you guys could point me in the right direction at the least (for example, would gearslutz be better? I'm not too aware of post-production sites).

Yes I know everybody has this pipe dream of using phase inversion to fix things, but that doesn't mean we can't try!

So, I'm doing sound for a low-budget reality show. Yes, it's trouble.

They wanted me to record sound in a noisy nightclub. Both boom and lavs pick up quite a lot of ambient noise from the DJ blaring the usual radio standards.

I am aware of noise reduction tools such as EQ, izotope, and cedar (probably out of our price range), but I was wondering if anything could be done with "phase inversion" stuff.

I only say so because the songs that are played in the club are available in a pure version at some place, are they not? One of the theories of phase inversion is that you record the original sound as clearly as possible so that it can be mixed in with the original track and thus canceled out? I mean a convolution wasn't possible to record (not that I ever recorded one in my short sound career) but could such a thing be done to the original track and have it mixed in with the production track?

In other words, try to emulate the acoustics of the club with the original track off the CD and mix it in with the production track?

Just to make it clear: This is to fix tracks already recorded. This isn't phase inversion on-location.

So am I dreaming? Is it possible?

Many thanks,

Sawrab

[soundslikejustin]

Yes, you are dreaming. Even if you recorded a convolution and applied it to the original, it would not quite work as I assume your subjects are moving, thus changing the phase/reflections that the mic is picking up. And it definitely wouldn't work with just applying reverb to original music in hopes of emulating a club full of people that are moving.

You would probably have better luck just taking a part of the audio from the footage that DOESN'T have them speaking in, and try phase inverting that. Again, it would need to be a similar part of the song, before the people have moved somewhere else to have the best results, but it might work.

EQ all the bass out, and use a multi-band expander to help yourself out.

Izotope offer a 15 day trial on their awesome RX software that you could utilise...

[rb1138]

Yes, you are dreaming. Even if you recorded a convolution and applied it to the original, it would not quite work as I assume your subjects are moving, thus changing the phase/reflections that the mic is picking up. And it definitely wouldn't work with just applying reverb to original music in hopes of emulating a club full of people that are moving.

You would probably have better luck just taking a part of the audio from the footage that DOESN'T have them speaking in, and try phase inverting that. Again, it would need to be a similar part of the song, before the people have moved somewhere else to have the best results, but it might work.

EQ all the bass out, and use a multi-band expander to help yourself out.

Izotope offer a 15 day trial on their awesome RX software that you could utilise...

Thanks for the reply. I'm not really a post guy. Can use all the help I can get.

Hmm they may be, but in some sections it's a sit-down interview. Do you think there's any hope for that? They do still turn their heads.

Yes I better get to work with the usual tools.

Best,

Sawrab

[soundslikejustin]

The sit down section might work, but again, it's probably not just the music (that you can't convolute and cancel out) but also other people talking OVER the music etc. It really needs to be a constant sound like a hum or buzz (like 60hz cycle from a ground loop etc) for phase cancellation to really have a good effect.

The head turns wont matter much in terms of changing the background seeing as they're wearing lavs, which pick up everything with an omni pattern.

What you're dealing with is quite broadband, but bass heavy. EQ out the bass, and then either use a multiband expander or grab the demo of Izotope.

There's a reason the club scenes in shows like Jersey Shore have subtitles...

[rb1138]

The sit down section might work, but again, it's probably not just the music (that you can't convolute and cancel out) but also other people talking OVER the music etc. It really needs to be a constant sound like a hum or buzz (like 60hz cycle from a ground loop etc) for phase cancellation to really have a good effect.

The head turns wont matter much in terms of changing the background seeing as they're wearing lavs, which pick up everything with an omni pattern.

What you're dealing with is quite broadband, but bass heavy. EQ out the bass, and then either use a multiband expander or grab the demo of Izotope.

There's a reason the club scenes in shows like Jersey Shore have subtitles...

Thanks! That's awesome.

The club scenes in Jersey Shore have subtitles? I've actually never watched the show. Not really a fan of reality tv.

[soundslikejustin]

The club scenes in Jersey Shore have subtitles? I've actually never watched the show. Not really a fan of reality tv.

Don't. It's horrendous. But yes, when there's something being said that the audience 'needs to know' and its buried under thumping club music, they often throw subtitles up.

[Jay Rose]

+1... we don't have a trick yet that'll help in your situation.

Maybe as voice recognition gets better (Siri and her sisters driving that), it'll couple with speech synthesis and we'll be able to enhance or artifically replace* this kind of dialog. But this week... sorry, nothing.

--

* I'm guessing you don't have money for ADR.

[studiomprd]

" In other words, try to emulate the acoustics of the club with the original track off the CD and mix it in with the production track? "

of course you are free to try it...

once again: unreasonable expectations.

[Bondelev]

You are assuming an awful lot. The playback of the music in the club will almost certainly ot be at the exact same speed as the original record and will ot remain phase-coherent throughout. In addition, the EQ of the speakers and the room reflections will not phase properly either. If you were to try this, you would need to mike the speakers in the room and multitrack it. Even then, it won't work.

RX even at its best cannot work miracles and makes the dialogue sound thin and aliased.

Your best bet is to explain all this to the producers and point out that if the music remains audible they will need to buy the rights to whatever is playing in the bar or the footage cannot be used. That will probably force them into rethinking this decision.

[Henchman]

You are SOL.

As are the producers. As they can get sued if any of the music is recognized, and not cleared.

[Philip Perkins]

As is so often the case in our world, the real fix for this issue is not technical. It is in the conception of the show and the choice of locations.

phil p

[Jesse Flaitz]

Does anyone else hear these stories on set of someone (who more often then not isn't even a sound person) fixed some CRAZY noise by phase reversing and playing over the dialogue... I've heard this a few times now when somene is trying to justify to me recording a scene that I tell them will have BG Noise rendering the takes unuseable.

[rb1138]

Does anyone else hear these stories on set of someone (who more often then not isn't even a sound person) fixed some CRAZY noise by phase reversing and playing over the dialogue... I've heard this a few times now when somene is trying to justify to me recording a scene that I tell them will have BG Noise rendering the takes unuseable.

Yes I've heard the story. You can see the result above: the original post, hahaha.

Nah, but I am trying to recreate such a story right now.

Best,

Sawrab

[Philip Perkins]

A really long time ago I got it to work once, sort of, because a video engineer on the same job told me about this "trick". The result was strange sounding, with a bit less BG. Since this video engineer considered himself the "engineer in charge" and "technical director" of that job I did it his way, and never heard from post. Any other experiments I've tried with this were utter failures. When I've been asked about this technique more recently, I usually respond: "Have you actually ever heard what that sounds like?"

phil p

[Marc Wielage]

I am aware of noise reduction tools such as EQ, izotope, and cedar (probably out of our price range), but I was wondering if anything could be done with "phase inversion" stuff.

iZotope RX really works best with steady-state background noise like traffic rumble, low-frequency hum, air conditioning wheez, or a high-frequency squeal from a light. Nightclub background music goes all over the place in terms of volume and frequency content, so it'll be impossible to chase. I have used combinations of Countryman B2 directional lavs and iZotope to tame very difficult interview situations, but it's a compromise, no matter what you do.

Note that iZotope RX2 is only $349, so I would hope it was within your price range. Realistically, what the reality producers are asking for is not doable. There might be a compromise, like moving the people to a private room with controllable volume, but if it costs $10,000 to clear every song fragment that bleeds through, this could be a very, very expensive proposition.

[rb1138]

iZotope RX really works best with steady-state background noise like traffic rumble, low-frequency hum, air conditioning wheez, or a high-frequency squeal from a light. Nightclub background music goes all over the place in terms of volume and frequency content, so it'll be impossible to chase. I have used combinations of Countryman B2 directional lavs and iZotope to tame very difficult interview situations, but it's a compromise, no matter what you do.

Note that iZotope RX2 is only $349, so I would hope it was within your price range. Realistically, what the reality producers are asking for is not doable. There might be a compromise, like moving the people to a private room with controllable volume, but if it costs $10,000 to clear every song fragment that bleeds through, this could be a very, very expensive proposition.

Yes iZotope RX2 is the idea. I meant Cedar was out of the price range. Either way, I have had success with iZotope in the past, but as you say the club music is quite varied. It doesn't seem like something iZotope can deal with. It's why I started the topic to see if there was anything else I could do.

Thanks everyone for all the help though. Awesome.

Sawrab

[Jay Rose]

How about using head mics? An E-6 is visible in closeups, but hardly distracting... and they can save the dialog.

Make sure you get a mic with enough headroom, if the talent will be shouting over the music. Countryman makes a low-sensitivity version for that.

[Philip Perkins]

Way back up there Mr. Henchman mentioned another roadblock--rights to the music. Worth mentioning if you think they haven't thought of this.

phil p

[JDirckze]

I did the phase invert trick a few years ago with good results. Recorded a womens choir with a backing track, then had them stay still and silent while I recorded the track by itself. We then laid it over a video of a live performance where the audio was no good. Worked rather well...

[Bondelev]

Way back up there Mr. Henchman mentioned another roadblock--rights to the music. Worth mentioning if you think they haven't thought of this.

phil p

Someone else mentioned it right above him.

[Jay Rose]

Jason, that is a cool trick! Obviously it won't work for the OP's situation, and you'd need an extremely speed-stable playback source and probably had the mics on stands so they wouldn't move... but definitely brilliant.

[rb1138]

I did the phase invert trick a few years ago with good results. Recorded a womens choir with a backing track, then had them stay still and silent while I recorded the track by itself. We then laid it over a video of a live performance where the audio was no good. Worked rather well...

Hmm could you do that with any live music played for an event? Assuming you make everyone stay still and the microphones didn't move....

Sawrab

[Marc Wielage]

Hmm could you do that with any live music played for an event? Assuming you make everyone stay still and the microphones didn't move....

No, the phase is going to change due to timing and acoustics, since the actors most likely are not going to sit down in the same space at the same time, plus there's going to be body movement and other sounds (footfalls, glasses clinking, people talking) that will prevent everything from cancelling out 100%. Even if you do try to cancel it out, what will happen is that the phase cancellation will modulate depending on the difference signal level as it rises and falls.

You can "chase" this to a point with natural-sounding background noise, like traffic rumble, waves crashing on the beach, or fan noise, but you can still hear the background noise level rise and fall due to the gating action of the notch filters. I hear this all the time in reality shows, competition shows, even dramatic productions shot under difficult conditions. Actor talks, you near the noise seem through; actor stops talking, noise stops. You can blend in background FX to try to put a bandaid on it, but it's a compromise at best, and I can't see it working with music and live nightclub noises.