The process of matching ADR recorded in a studio, with the production sound recorded on location, maybe 6 months previous, is a key role for the re-recording mixer. In this article, Garrard Whatley and Damian Kearns share their secrets on getting ADR to sit well in the mix.

Garrard Whatley

Producers hate it - unless they add it. Actors don’t like it much either. But ADR is an occasional begrudging necessity with so many variables that it’s honestly surprising how much of it ends up working out. And the pandemic definitely made things worse, with some actors recording at home on their phones!

Getting ADR to sit well in a mix - honestly and necessarily - starts at the beginning. It’s a long chain of events leading to hopefully the best outcome for dialog in the project.

1 - Create The ADR Cue List

There are two overarching reasons for ADR:

1. Production noises that step on the dialog

2. Adding or changing cues to help propel the story (or to change the story)

These can be dissected further, but really, that’s it.

An obvious example of ‘production noises that step on the dialog’ would be a plane flying overhead in a Napolean-era drama but could be any on-set noise, or a director’s voice cueing the actors.

Adding cues as in ‘to help propel the story or change the story’ may come in the form of additional dialog when an actor is off-screen to help explain a storyline that maybe was confusing to test audiences, or to accommodate a plot change covered in a re-shoot.

On any film or TV show, one of the first things that needs to be done, after spotting the show with the producers and director, is to create the ADR cue list. I ask them to provide any additional cues or changes that they want (which is sometimes covered by temp dialog or on-screen text in the AAF and QuickTime) in order to add them to my master ADR list.

There are great tools now for correcting a lot of wrongs in production dialog, helping to shrink ADR cue lists. But since producers are usually anxious to get the list quickly, it can be better to spot every potential ADR cue and deliver that, with the caveat that it will look smaller soon; and then later, go through the cues to see what can be eliminated from the list. Otherwise, tinkering with dialog while creating the list may take too long to deliver it in a timely fashion.

As experience grows, an editor will have better instincts to know what belongs on the list, and what tools she has to address a particular type of problem.

2 - Reduce The ADR Cue List

For me, this must include iZotope RX Advanced. With so many modules, it is now a one-stop-shop for saving dialog. I’ve been able to remove huge slamming sounds, crickets, birds, encroaching footsteps and other noises, in addition to simply lowering the noise floor.

Precision and speed increase with experience and research, find what others are doing to remove unwanted noises either online at sites like this one, or by visiting product demos or industry conventions. Sometimes a module created for a specific purpose might work great for addressing a completely different type of problem!

With the list whittled down, and delivered to a grateful (maybe) producer, it’s time to schedule actors.

3 - Record The ADR

If possible, I assign myself or a very trusted mixer to record the ADR. We know best what mic to choose, placement for that mic (and lavalier), whether the actor was speaking from the diaphragm or more from the chest or head, and other fine details such as pitch. These specifics will help tremendously with matching ADR.

But it’s the actor that must deliver. And that is another entire set of variables. All have different levels of experience with ADR. Perhaps their mind won’t be in the game because they drove through bad traffic on the way to the studio, or have a sick relative.

A lot of things can affect performance and endurance. That’s one reason why it’s important to have as short a list as possible.

I always suggest that mixers take some time to practice performing ADR themselves. Get in the booth, listen to the beeps (or pips), and try to match the dialog for multiple lines. Do it until it’s done right. Empathize with what the actor has to go through mechanically and emotionally. Then you can speak to them with clarity and encouragement, and provide an atmosphere that helps them focus on the task at hand and maybe forget about that bad traffic, for just a bit.

I’m sure we all have stories, some just awful, like an actor who is clearly high and slurring words after arriving an hour late, or an actor who keeps walking off-axis, etc., and then some who are just fantastic. So, we control what we can, giving little tips along the way if necessary, to record a line of dialog that just might work in the end.

One actor at a time, one cue at a time.

4 - Prep the ADR

This step begins with editorial and EQ.

Editorial - In ADR, for a long time, the Synchro Arts plugin Vocalign was a big name in compressing/expanding ADR cues to match source dialog. You supply the plug-in with a ‘guide’ clip and a ‘dub’ clip, and it returns an altered dub clip that matches the waveform of the guide, with varying levels of user-controlled flexibility.

I always felt the results were mixed at best. More often I would have to trash the newly created clip because there would be too many digital artifacts, or at best keep only a portion of it. Occasionally it would hit gold though and help out tremendously. Otherwise, I would simply do it manually, expanding clips with existing plug-ins like Serato’s Pitch ‘n Time or zplane’s Elastique.

Synchro Arts have since developed Revoice Pro 4 which works much better, with the same guide and dub clip workflow. While you always have to be wary of artifacts, the results are much better and can be used much more often. Additionally, it gives you the option of matching the guide track by pitch as well as timing. I have seen ADR that was marginally good become ADR that matches much more perfectly in a scene when using the pitch function. But don’t get too carried away with it, artifacts may abound!

Pro Tip: running some noise reduction on the guide track may help steer the algorithm in the right direction for matching syllables. Sometimes the software may view noise as dialog and give you some bad matches.

When it comes to ADR editing, keep in mind that many times you can use only what you need -- so, for that background sound that slammed over only a syllable-and-a-half, try using only the frame or so you need from the ADR rather than replace the whole line. It may do the trick better than having one full sentence that sticks out like a sore thumb.

And to be perfectly candid, I’ve supplied my own voice to fix dialog problems when nothing else would work. I am 100% certain that I am not the only one who has recorded a “ch” to finish a sentence, or even replaced full words, without telling the director -- and having no one ever notice!

EQ’ing the ADR - This can certainly be done manually quite effectively by creating a playback loop across a bit of original dialog on either side of the replacement cue. It can also get mind-numbing depending on the time of day!

iZotope has created an EQ Match module that, much like Revoice Pro, requires a guide and dub clip, plotting the general EQ of the original dialog, and applying it to the ADR clip. For me, at worst it has been a great starting point and sometimes is spot-on.

As with Revoice Pro 4, running some noise reduction on the guide clip may be handy in sorting the dialog EQ from the background noise EQ. Keep in mind that with iZotope you can select just the dialog portions of the clip, the ones with the character needed speaking, and not other characters off-axis, to give the algorithm a chance to determine the correct EQ settings.

5 - Mix the ADR

Once you have got the ADR matching the surrounding dialog fairly well, it’s time to give it some life using reverb, level, and possibly more EQ’ing and compression. First, add some mono reverb to match the surrounding dialog, again you can create a playback loop and change reverb levels on your sends to get it right, but over time it will become second nature. I like to use Altiverb, but everyone has their favourite and I wouldn’t begrudge those experiences.

As mentioned in another article, How To Use Reverb In Drama Production - Expert Panel, I have both mono and stereo sends for reverbs. The mono in this case is used to have the ADR match the surrounding dialog clips, then a stereo or surround send will be used across the master dialog aux for the full sound of the room.

Usually, some leveling, EQ’ing or compression will take place to set it in the scene a bit more. A little extra compression can really help with some ADR that sounds a bit too crisp.

I’ve noticed as well that ADR often needs to be just a little quieter than we think on that first pass - much like foley, try bringing it down a little and see how it feels.

6 - Trash the ADR

You get to the Final Mix! and the director hates an ADR line read, but the producer loves it. The mixer then gets to sift through the bickering to try to come to some middle ground. How about the first half of this take with the second half of this other one? And quickly go through the steps, if necessary, that Synchro Arts and iZotope have made it easier to get it done quickly.

And finally, as sometimes happens, have that original dialog handy. During the final mix, hearing all those backgrounds and fills - not to mention the music - sometimes the dialog we thought was awful ends up working the best. So, trash the ADR! The best performance was right there all along. The director is happy, the producer is happy, and neither one knows the amount of work you put into the ADR. But that doesn’t matter…

Time to move on to the next cue!

Garrard Whatley owns Seismic Post Audio in Los Angeles and has worked as a Supervising Sound Editor and Re-recording Mixer for 24 years. His first company RocketWerks was located in Santa Monica, primarily catering to independent films. Still working on indies and larger film productions, he has added television series work over the last 10 years, bringing a cinematic sensibility to his broadcast work. Seismic Post Audio is also home to software development for post audio applications, with an eye toward making Pro Tools session organization and navigation faster and easier.

Damian Kearns

In the earliest stages of my career, Automated Dialogue Replacement (ADR), was typically a task assigned to a junior engineer or in one singular case, an engineer who specialized in recording ADR lines. This approach has never made much sense to me since I believe mixing starts at the microphone and ends when I’ve successfully delivered my final mix audio files to the distributor or broadcaster.

Without enough knowledge of the challenges of editing and mixing post-production audio, an ADR recordist might, and often does, deliver less than ideal material. In the case of ADR, perspective is everything, so here’s a little perspective.

The first time I assisted an ADR session for a truly great recordist was in 1996 in Vancouver and he also happened to be the facility’s head engineer. His holistic approach to audio imprinted itself on me; from him, I learned there was no audio task too big or too small. Every element is fun to work on and requires skill and creativity to execute properly. Mastery of technique is always the end goal.

Perspective Is Everything

We used a Sennheiser MKH 416 or some other similar shotgun patterned mic for the main mic but also set a Neumann U87 towards a far corner, using an omni polar pattern. He was the ADR Mixer and I was the Recordist. I operated the tape machines and wrote the take sheets. What I felt made his ADR really special was he was constantly playing perspective between the two mic’s, riding EQ and compression on the fly, and wasn’t shy about asking the talent to move around. And his levels to tape were perfect. This man had done a lot of ADR for Hollywood movies and big-budget TV shows. He was the guy to learn from!

We did sessions for various movies and a lot for TV shows like Millennium and X-Files. Over time, his technique became my technique, as it was clear when we were mixing shows using ADR that was recorded elsewhere, that the attention we were giving our recordings made ours much easier to deal with.

When I moved on to the Canadian Broadcasting Corporation in Toronto, I was tasked with ADR recording from time to time. At one point, I was even assigned a large share of the recording for a TV series.

I dealt with actors and loop groups (a group of people hired to sound like crowds of people or to record non-specific wild lines for background actors) on a weekly basis during that gig. I remember using the same techniques I learned in Vancouver: Playing perspective and balancing a close and a far mic to tape).

I also did some other stuff I like to think I originated (but probably didn’t!). I had lines for a character hiding under a desk so I had the actor hide under a desk and mic'd that perspective to match the guide audio I was using as a reference.

In another case, I had someone on the other side of a window speaking into a restaurant so I pulled up a baffle I had with a window in it and mic'd the actor from both sides, to give the re-recording mixer options. I was looking at my meters a lot to ensure healthy levels but mainly, as a person who was also a re-recording mixer, I was listening for a match. I remember the veteran dialogue editor who was in the room with me commenting that he learned how to record ADR properly by attending our sessions. He was there to ensure performance and lip-sync he could use and I was there, supporting with all the technical might I could muster. We had great success in those sessions. And I’ve had great success with ADR since then.

In my experience, there are six main aspects to matching ADR to Production Dialogue. They are…

1 - Mic Choice And Placement Technique

Getting the sound is crucial so working against the production audio, I get as close a match as possible. To this end, I have an MKH 416, a U87Ai, and a couple of different wired lavalier mic’s in my studio. I personally favour the 416 over lavalier mic’s, even when I’m replacing lav audio. I just love a good boom mic over a lav mic and experience has taught me how to play with my 416 to get it to sound right. I usually have a second 416 on another stand, attached to a 100 ft. cable, so I can use other rooms in my house to capture the right space when I need to.

I do use the lav’s when I need to get their specific perspective. I typically pin a lavalier mic, approximating the production technique, to get it sounding similar to production. Then I mute that record track so I can concentrate on the 416/U87Ai mix. I only keep my lav mic open when I actually have to match the lav with finesse.

2 - Lip Sync

This is best dealt with during the performance, as much as possible. I’d rather not have to run Revoice Pro or Vocalign Ultra too hard if I don’t have to so directing talent to carefully mimic timing means less or no post processing or editing required. This stage can be incredibly time consuming and in my opinion, it’s often unnecessarily so.

3 - Level

Level is the key to a good technical match. With modern EQ and Ambience Match software, the missing quotient is often lost when the ADR and the production sound are not set initially at or near the same level, prior to processing. If I apply EQ Match and then have to turn my ADR up in level for instance, the EQ curve naturally changes with gain. So level matching is key.

4 - EQ And Compression

EQ gets me closer to a match than anything else other than mic placement. Sometimes, a little compression helps here too. What I typically do is either manually EQ or run an EQ Match module or plugin and play with the percentage until I have what sounds like two pieces of audio in the same spectral neighbourhood.

5 - Reverb

Reverb doesn’t always have to happen but when it does, it must be mono. It must! Microphones are mono. Matching a mono production mic to a mono post-production mic requires a mono reverb. I have tracks in my template that are mono reverb tracks for precisely this purpose, as well as to smooth Franken edits. Match your space using (ideally) a convolution reverb setting that matches or approximates the scene, or use an algorithmic reverb you know extremely well and can tweak with finesse.

6 - Ambience

I add the ambience at the end. I sample and match the production ambience and play with the amount I add to the ADR until I can slot the ADR perfectly into the scene. To this end, I use either RX 8’s Ambience Match module/plugin, or my favourite dialogue toy, Dialogue Match.

In Conclusion

As much as ADR relies on mic perspective, it also relies on mix perspective. To achieve great ADR recordings, try to imagine yourself working with your recordings in the mix. The ideal is to record something that can, with little effort, sit in amongst production audio without calling attention to itself. We may not be able to control the talent but we do control everything else. And it starts with the mic and the recordist.

Damian Kearns is an extremely creative, versatile audio engineer. He specialises in post-production for television and movies and has many promotional, pre-production, production and presentation audio credits. This past winter, Damian has led the post-production team as lead mixer for "Fear Thy Neighbour," "Fear Thy Roommate," and "History Erased (season 3),' as well as the UHN podcast “Behind the Breakthrough (S2)”.