When editing dialog, I have found that a few key concepts are at the forefront of my decision-making process. I’m 100% certain I will miss some things, as I’ve never actually codified my thought process until now. Furthermore, this isn’t a checklist I go through on each decision. A lot of editing becomes rote. But when I have a difficult decision, or a major one that will impact the way I work on an entire scene, this is how I will proceed.
You’ve all heard of the triangle of choices: a client can usually pick two: “cheap, easy, and good.” I find that difficult choices in dialog editing often follow a similar route. Great production sound allows you to have all four of these:
- Consistency
- Intelligibility
- Sound Quality
- Ease of use
However, as production sound becomes less and less optimal, you may end up only being able to pick two. In the worst case scenario, only one! Better hope there’s an ADR budget for that!
These picks aren’t “all or nothing,” though. They’re more like balancing faders. At any given time, making a choice may increase one pick and decrease another (or even multiple), and not always by the same amount!
Below I’ve defined each and given some discrete decision types that could affect each picks score. Notice that there is overlap, which is why sometimes there will be trade-offs.
Consistency
A scene should be able to play, front to back, with no considerable (or, importantly, perceivable) changes in sonic character. This includes foreground voices as well as background noise. If a scene fails to play consistently in the edit bay, I haven’t done my job. You can maintain consistency with some or all of these techniques:
- Cutting from shot to shot, or character to character near the beginning of a line. Temporal masking will hide minor changes in tone
- Long fades on noisier shots can make the added noise less noticeable as our ears adjust to it
- Adding fill underneath a clean take when the surrounding takes aren’t clean
- Opting to use “only boom” or “only lav” for the duration of the scene where possible, providing fill when it’s not possible.
Intelligibility
Most foreground dialog is meant to be heard and understood. Dialog is, in western cinema especially, the single most driving force for the narrative. In my biased opinion, more-so than any element of picture! (You can consume a film fairly easily without watching it, but try watching most films without sound or captions and they fall apart) There will be times when intelligibility doesn’t matter, of course: made up languages, mumbling that no one on screen can understand, efforts, etc may not need special attention to intelligibility. But as a general rule, this concept is absolutely vital to every sound a character vocalizes.
- Selecting mics with minimal noise is, of course, hugely important. Lavs can be great for this in tricky locations, if the PSM and costume designer did their jobs well. However, if the frequency response is out of balance, the extra eq added in the dub could bring the noise floor up unexpectedly. If I suspect that might be the case, I throw an EQ on the track and do a quick balance, just to hear whether it’s actually usable. Obviously, I take the EQ off immediately! The mixer doesn’t want that in their session!
- Selecting mics with a balanced frequency response, even if there’s a bit more noise, is often preferable as it will sound more natural and therefore be more intelligible. Often, this means you’ll be selecting the boom mic, as lavs tend to be heavy on the low end.
- Removing off-mic dialog. If you are running two lavs open, you run the risk of weird phasing, more room reverb, and other unpleasantness. Either cut between the mics or run fill instead of off-mic dialog. If you can’t, for whatever reason, applying Auto Align Post to dynamically resolve phase issues between the mics will significantly improve the result.
- Related to off-mic is overlaps. Repair overlapping dialog (where one actor steps on the other’s toes) if possible with takes where they didn’t overlap. This way the mixer will be able to choose the best mix. Remember to watch the noise floor with a repair like this!
- Removing unwanted production sounds: camera dollies, light switches, background props, sometimes foreground props. If they’re between lines, just kill them with fill. If they overlap with dialog, look in alternate takes for a syllable that’s un-dirtied, or use spectral editing to remove it, if it’s easily removed that way. Just remember to either make a muted copy of the original, underneath, or only edit that small portion so the handles of the original are still available in case your work was… overzealous.
- Letting both boom and lav play simultaneously. Some mixers want this. Some dread this. Sure, it’s double the work for both you and the mixer, but if the boom is too roomy and the lav is too boomy (I’m sorry, I had to), there may be no other way to achieve intelligibility! Again, Auto Align Post is your best friend in this case.
Sound Quality
This concept is a bit nebulous. It’s kind of a “you’ll know it when you hear it” characteristic, and frankly, could probably be wrapped up in consistency and intelligibility. I’m giving it it’s own concept slot because sometimes, you can achieve consistency and intelligibility in multiple ways, so you have to make a decision on how to get there.
- Broadband noise
- Production noises
- Frequency response
- Airplanes
- Car honks
All of these and more can affect sound quality. You all know this. Why am I telling you?! Just so I can have pretty bullet points in this section? Probably.
Ease of Use
“But film production isn’t supposed to be easy!” you say, while waving your hands like a mad-person. True, but steps can be taken to ensure that work is easier. And often better because of it. Any time you can find a way to get the equipment out of the mixer’s way (DAW included), you’ve made the project better, because they can just focus on mixing.
- Checkerboarding. Obviously. Don’t even consider not doing it. I didn’t checkerboard for years and, well, I’m only just becoming good at my craft at 32 years old because I got complacent. If you don’t know what this means, Google it, or read “Dialog Editing for Motion Pictures” by John Purcell.
- Part of this is splitting up tracks. Whoever is mixing your show will have opinions on this. Check with them. Consider splitting a loud yell in the midst of a whisper so the mixer can place the fader instead of ride it. That one moment that has a lot of noise compared to the rest of the track? Split that out. Actor is off mic for one word and you can’t for the life of you find and alt take? Maybe you want to split that. Someone laid on the horn for 30 seconds during the only take? Split. (Yes I’m speaking from experience!)
- Selecting only one mic if possible. Again, another detail a mixer may or may not veto. Listen to the mixer at all costs. If a scene clearly sounds great with just a boom, put the lav in a Junk or X track and just keep going. If you’re the mixer and you KNOW 100% you’re not gonna use the lav because it’s terrible and the boom is great, trash it. Or vice versa. When I mix, I vastly prefer one mic over two as long as the mic chosen sounds good.
- Remove useless regions. If they aren’t contributing to the scene they obviously need to go.
Here’s the thing about these picks. It’s not about balancing them. You want to, if possible, max them all out! That’s rarely possible, of course, so you have to decide on a moment-by-moment basis which parts can be sacrificed and which can’t.
For one scene, I may find that consistency is the most important pick because it’s a super intimate, natural scene that’s going to be played devoid of any sound effects or music. Any changes in the sonic space will be multiplied.
In another scene, it’s a busy basketball game with no important dialog except for one or two lines. Decisions can be made based on sound quality and ease of use given the wild rock music, crowd, and sfx you’ve been told by the spotting session will be present. Except those two lines, which will, ideally maximize intelligibility. Still, because of the music and crowd, consistency is less important here! (That doesn’t mean ignore consistency. You’re boned if the mixer decides that those two lines need to exist in a vacuum, and they turn down the music and FX VCA’s only to hear that you cut between boom and lav liberally within the line to make it “intelligible”)
Again, this is just a framework. Not a checklist. In the end, as everyone says, “there are no rules. If it sounds good it is good.” But this is a way of reasoning through tough choices, and invariably I come back to optimizing these four things every time.
Thanks to my mother, sister, brother, and wife on deciding what to call this post!