Jump to content


Photo

A simple question about ADR.


  • Please log in to reply
4 replies to this topic

#1 Rob Thomson

Rob Thomson
  • Basic Members
  • PipPip
  • 18 posts
  • Director

Posted 05 December 2011 - 02:49 PM

Hello,

There's a program called Pluraleyes (http://www.singulars...luraleyes.html ) that's used for when you're recording dialogue to a source that isn't the camera. It 'listens' to the poor quality sound coming from the camera's own microphone, and replaces it automatically with the good sound recorded from your sound recorder, saving hours of synch editing.

I had to dub/ ADR all my dialogue in a film I'm making, and of course, it's taking a while matching every single line of ADR'd dialogue with what's on screen.

My question is this, would plural eyes be able to synch ADR'd audio over the original recordings?

To be honest, I'm guessing it wouldn't work, so I don't want to spend $150 on something that's going to be totally useless to me, but of course I could be wrong, and pluraleyes could be a lot smarter than I think.

Could someone give an educated guess, or if you know it will/won't work, please reply!

Thanks,
Rob.
  • 0

#2 Chris Millar

Chris Millar
  • Basic Members
  • PipPipPipPip
  • 1642 posts
  • Other

Posted 06 December 2011 - 01:22 AM

disclaimer: I don't know anything about the pluraleyes algorithms - but I might be able to make a partially educated guess... (at least start the conversation Posted Image)

<guess>
no
</guess>

I'd say it works somewhat like shazam et al. in that it's looking for fingerprints in the frequency/time domain (as opposed to the amplitude/time domain) - looking at the audio 'sideways' in some regards. It likely finds the track with the greatest dynamic range i.e. the mostest well recorded sound and creates an image of which to compare the the lesser quality recordings with (the on-camera recordings).

But the nature of ADR is that it is different, the image from it will just ... well, be different - because, um, it's different ... If ya get what I mean. Posted Image

Maybe something smarter as you suggest would be an implementation of voice recognition Siri like algorithms. Or maybe something as simple as a low pass filter on the ADR, just work on the really blunt information to get you in cooee. In the frequency/time domain this would be the equivalent of decreasing the image resolution as opposed to truncating the top of the image off or blurring it (which were my initial thoughts).

But then if you're recording ADR wont you be using the footage in the first place ? Wont it be synced by the very process of recording it already ? Maybe I'm showing my ignorance of the processes involved.

http://www.ee.columb...ng03-shazam.pdf
  • 0

#3 Brian Drysdale

Brian Drysdale
  • Basic Members
  • PipPipPipPip
  • 5070 posts
  • Cinematographer

Posted 06 December 2011 - 04:37 AM

But then if you're recording ADR wont you be using the footage in the first place ? Wont it be synced by the very process of recording it already ? Maybe I'm showing my ignorance of the processes involved.


It seems like he has recorded a poor quality guide track using the camera's on board microphone. I assume this is a video camera

For the amount of bother involved, I'd tend to record good quality dialogue at the time of shooting, unless you're in a noisy environment that doesn't allow satisfactory recordings.
  • 0

#4 Tim O'Connor

Tim O'Connor
  • Sustaining Members
  • 860 posts
  • Other
  • Boston, Massachusetts

Posted 06 December 2011 - 04:09 PM

Just sync it the way that everybody does anyway when you record audio to match something that was shot previously. Put it on your timeline and move it around to get the best match. It's unlikely that it will ever sync in any true sense but it can match up well enough so that people will never notice, if you get good audio quality and good line readings. Programs like you mentioned are for synchronizing recordings from double-systems that were rolling at the same time. For ADR you are recording picture and audio at different times.
  • 0

#5 Marcus Joseph

Marcus Joseph
  • Basic Members
  • PipPipPipPip
  • 404 posts
  • Other
  • Sydney, Australia

Posted 06 December 2011 - 11:08 PM

Syncing dialogue is a difficult talent to master, of course with more money you can throw at someone to do it, the better it will be, unless you find some extremely talented sound guys in a film school.

I've seen good dialogue editors work and they can rip back the frequencies and crossfade what they don't need to get a match on the original recording, it takes time and practice to get good at it, but no program can do it. That program's only for syncing double system sound with on board recording and Final Cut Pro X can now do that automatically too. But you yourself have to try to match up the waveforms as close as possible anyway.

You must also learn how to EQ the sound and what have you.
  • 0


Paralinx LLC

Abel Cine

FJS International, LLC

CineLab

Tai Audio

Glidecam

Wooden Camera

Visual Products

Metropolis Post

Aerial Filmworks

Gamma Ray Digital Inc

Rig Wheels Passport

rebotnix Technologies

Willys Widgets

Opal

CineTape

Broadcast Solutions Inc

Technodolly

The Slider

Media Blackout - Custom Cables and AKS

Ritter Battery

Media Blackout - Custom Cables and AKS

Rig Wheels Passport

Tai Audio

Ritter Battery

Metropolis Post

CineTape

rebotnix Technologies

Abel Cine

Paralinx LLC

The Slider

Wooden Camera

Visual Products

CineLab

Broadcast Solutions Inc

Glidecam

FJS International, LLC

Aerial Filmworks

Opal

Technodolly

Gamma Ray Digital Inc

Willys Widgets