Dialogue Editor

What they do

A cut assembled from many takes has audio that jumps: different mic positions, different room tone, a plane in take three, an air conditioner that came on halfway through the day. The dialogue editor makes it one continuous performance. They open the production sound in Pro Tools with every recorded channel available, boom and radio mics, and choose per line which mic and which take gives the clearest reading, sometimes lifting a single word from an unused take and fitting it inside another. Noise is removed with spectral repair and broadband tools rather than blanket denoise, because over-processing hollows out a voice. Room tone fills the gaps so the background does not pulse at every edit, and perspective is matched so a wide shot does not sound like a close-up. Where a line is unusable or the performance changes in the edit, the editor writes ADR cues with timecode and text, runs or attends the ADR session, then fits the studio recording to the original lip movement and treats it so it belongs in the location. The session is then checkerboarded across tracks, each character or perspective on its own, labelled and handed to the re-recording mixer, along with a clean split that has no non-dialogue content in it because platforms require a dialogue stem and a separate effects-and-music pass.

What they decide

The calls that belong to this role and no other.

• Which recorded mic and which take carries each line, including word-level substitutions
• How much noise reduction a line gets, and where the limit is before the voice degrades
• Which lines need ADR, and the wording and timing of every ADR cue
• How room tone and fill are used to hide edits and level changes
• Track layout and checkerboarding handed to the mix
• Whether a scene's dialogue problem is fixable in editing or needs to be flagged as a reshoot or full ADR pass

Principles of the craft

• Intelligibility first, then continuity, then character. A line the audience cannot parse has failed regardless of how good the take was. Once it is clear, the job is making it match the lines either side of it.
• Fill is what hides an edit. A dialogue cut is audible because the background changes, not because the voice does. Matching room tone across an edit makes the edit disappear without touching the performance.
• Perspective is information. A voice recorded on a radio mic on the chest sounds close no matter what the shot is. Matching mic choice to shot size is what keeps sound and picture describing the same space.
• Denoise is a trade, not a fix. Every decibel of noise removed takes something off the voice as well. Two lighter passes with different tools usually beat one aggressive one, and cutting to a cleaner take always beats both.
• The dialogue stem has to survive being taken away. Because localised versions replace the dialogue and keep everything else, anything non-dialogue hiding in the dialogue tracks will vanish in every dubbed territory.

Common mistakes

• Blanket denoise across a whole scene, producing a clean, thin, underwater voice
• Cutting between boom and radio mic mid-line, so the voice changes character inside a sentence
• No fill, so the background drops out audibly at every dialogue edit
• Writing ADR cues from memory rather than from the exact words in the take, which wastes stage time
• The experienced editor's mistake: fixing a line so thoroughly that it no longer matches the imperfect world around it
• Leaving effects, breaths from other characters or music bleed in the dialogue tracks, which breaks the M and E deliverable

How to learn it

Exercises and routes in, not a reading list.

• Record the same monologue three ways: boom at correct distance, radio mic on the chest, and a phone across the room. Cut them together mid-sentence and listen. Then try to make the phone take match the boom and note exactly what you could not recover.
• Record thirty seconds of room tone in a space, then record dialogue there. Cut the dialogue into a different scene using only that room tone as fill, and check whether the edits are audible on headphones at high level.
• Take a scene with traffic or air conditioning under it and clean it three times: once with spectral repair only, once with broadband denoise only, once by replacing lines from a cleaner take. Compare the voices, not the noise floors.
• Write ADR cues for a two-minute scene with timecode in and out and exact text, then have someone record to picture from your cues alone. Every place they had to stop and ask is a badly written cue.
• Build a proper dialogue split with nothing but dialogue in it, then mute it and confirm that the remaining tracks still describe the scene. That is the M and E test.