Post Scripts
Dialog Editing in Theatrical Films and Television
By Dane Davis

Movies aren’t reality. But movies must offer cross-sections of reality to their audiences. To succeed, there are a few simple rules of aural reality that cannot be broken: First, people speaking in the same room must sound like they are in the same room. Second, continuous time must sound like continuous time.

The illusions of real time and space are critical to screen storytelling. But the methods used to capture narratives on film naturally work against those illusions, because the time and space on a movie set has no relationship to the ultimate screen time. These illusions can only be maintained if one of movie-making’s least-appreciated crafts is executed flawlessly. These illusions are created by dialog editors, ADR editors, and the re-recording mixer.

Mind the Gap
When a movie is filmed, practical requirements determine the order of scenes to be enacted and captured. Multiple takes of each line are recorded from various angles. When all of this non-sequential information comes out the other side of the picture editorial process, the sound track will contain a lot of secondary, usually unwanted noises that contradict the screen illusion of time and place.

“Even a small gap in
an otherwise continuous
noise can destroy the illusion
of aural reality.”
Near any set, there are many sound sources that work against the illusion of continuous time in a consistent space. As the picture cuts from a shot made on Tuesday at 3:00PM to a shot made earlier that morning, or the previous week, or six months later, the background sounds on the set can be completely different. Even if the actor’s vocal performance feels completely seamless between the third take close-up and the take twelve two-shot (a camera angle that includes two characters rather than the single actor in the close-up), the jet in the distance or the truck idling at the intersection outside the sound stage colors this illusion.

When you cut between the two tracks, the sound “bumps” when the incidental noise stops or starts. This sudden change violates the illusion that a conversation is unfolding in a natural way right in front of the audience.

Dialog editing protects this perceived continuity. When noises cannot be removed in the editing process, the dialog editor must find a way to extend those noises across picture cuts in both directions, so that texture and timbre changes do not correspond to visual camera angle changes. For example, an idling truck noise behind the first line of dialog can appear to “drive away” by fading out during the second line (culled from another take recorded without the truck noise).

Adding wind noise to a quiet part of a windy take adds
continuity and builds ambience.
Even a small gap in an otherwise continuous noise can destroy the illusion of aural reality. So we smooth out large or small gaps by propagating what is around it, just enough to take the surprise out of it. This process is called “filling” the production track. Sometimes the offending noise must be extended across an entire scene to avoid calling attention to itself. The listening audience only notices these background sounds when they change, especially when they’re focused on the foreground of the actors speaking to each other.

A Mass of Distractions
Whenever possible, continuous obtrusions of this sort are edited out and filled — or “backfilled” — in the same way that discontinuous shots are smoothed out. Car horns, doors, crew noises, camera dolly creaks, microphone bumps, dog barks, and an evil cornucopia of other aural distractions can be replaced with a common background blur.

Actors contribute their own distractions as well, created by their limbs or vocal apparatus. Physical noises include chair scrapes or rattles, foot scuffs and steps, and setting down or fiddling with props over lines. What plays visually as character gesture often works against vocal expression if it dominates the audience’s ear for a moment. Anything that “takes me out of the movie” requires a period of watching time to dissolve back into the story.

“Sometimes an offending noise
must be extended across
an entire scene to avoid
calling attention to itself.”
Many actors, especially those who perform at a quiet level regardless of the sonic context, make a menagerie of mouth, tongue, and throat noises that sound like clicks, pops, and clucks. When the re-recording mixer raises and compresses the dialog to a comprehensible level against the background and action sounds, these inadvertent vocal noises become huge. The dialog editor must cut out these very brief but intense events. If they are literally too short to edit out, then he or she smoothes the waveform with the Pencil tool in Pro Tools.

When dramatically intrusive noises are right on top of words or phrases and cannot be segregated from the actors’ voices, the dialog editor looks for alternate syllables or words recorded during other takes to replace the problematic reading. When actors “step on” or clip each other’s lines and the overlap doesn’t work in the picture cut, replacement surgery is also required. If the same defect exists in the other takes, a more severe alteration is required. Sometimes “cheating” a syllable or a whole word from another scene entirely can rid the performance of intrusions like a bottle clunking down, a chair scraping as the actor stands up, or an off-screen door closing. Whenever possible, these alternates can be sent to the cutting room for a trial run in the cut done in the Avid picture editing system.

Splitting and Sculpting
A dialog editor essentially takes a scene full of very raw audio behind a carefully sculpted performance and prepares it for the dialog re-recording mixer. The mixer can use tools like EQ, notch filters, noise reduction, and gating, as well as level balancing, to create the final illusion of continuous on-screen time and space. This editorial preparation is critical for the mixer to work deftly and efficiently.

Split out a useful production sound effect and fill in the
hole with ambience.
The editor organizes the regions of audio into tracks that have common background sound characteristics. This process is called “splitting,” because it involves literally splitting an assembly of contiguous production track segments into groups according to timbre, rather than any dramatic logic. Any chunk that has a refrigerator compressor running through it gets joined together on the “A” track. What doesn’t goes on the “B” track, and so on. The editor can then extend the A track with “fill” that includes the fridge across the spots where the featured dialog on the B lacks it. It has to be mined from spaces between lines, pauses, or the rich but brief resources after “action” or before “cut,” when nobody is speaking.

Extending a buzzing noise can help ensure continuity.
This fill is often difficult to acquire, since it has to be recorded from the same mic position with the same acoustics and ambience as the lines themselves on the A track. If lights are shut off, or the crew or cast shifts position, or the mic boom points differently, then the ambience will not match enough to be usefully homogenous. Filling and extending only work if everything but the vocals matches closely — so when the shooting crew actually takes a moment to record “ambience” or “presence” on the set, the result is rarely usable, and is often just a recording of the set crew setting up for the next shot. But we appreciate the attempt, and sometimes the result can be a critical improvement over nothing.

Keep Breathing
Another key aspect of the dialog editor’s craft is the reconstruction of natural breathing by the actors. When the picture editor makes decisions about where to change angles with corresponding audio, a performer’s natural pattern of breaths between lines or words often gets butchered. Perhaps another character in the scene speaks, or the switch to a different take changes the pattern of inhaling and exhaling. The sound editor then must extend the pieces on his or her newly split dialog track to restore the original breaths, or replace what is no longer usable with a “cheated” breath from another take (or a completely different scene if necessary). Again, the illusion of reality can easily be broken by “chopped” breaths that have no natural rhythm or connection to the spoken lines. In fact, an actor’s breathing is often an integral part of the emotional performance, and needs to be respected and preserved as much as possible.

If the defects in a piece of dialog cannot be overcome by the dialog editor or the re-recording mixer, and is still too unclear or distracting, then the actor must be brought in to record the line again in a more controlled environment. This process is called ADR (Automatic Dialog Replacement) or “looping.” My next column will explore the world of the ADR supervisor, editor, mixer, and actor, and their contributions to this last step in making the final mix as natural and clear as possible. All of these crafts are about maintaining an invisible structure underneath that expected illusion of continuous time and space.