Almost everything written about sound and picture editing treats them as consecutive. You cut, you lock, sound starts.
That is true of the schedule and false of the work. By the time a cut is locked, a set of decisions has already been made that determine what the mix is able to do, and most of them were made for good picture reasons without anyone thinking about audio at all. None of what follows asks you to edit differently for the mixer's convenience. It sets out what each choice costs downstream, so the trade is visible while you can still make it.
Picture Lock Is a Promise About Structure, Not About Taste
Picture lock means the timing and the order of shots are settled. It does not mean nothing can change: colour, titles, VFX shots and graphics all continue to land after lock without disturbing anything.
The distinction that matters to the audio team is whether a change moves frames. Reordering shots, trimming a shot by two frames, extending a scene, inserting a beat: all of those shift every subsequent audio event on the timeline, which means work already built has to be repositioned. Swapping a VFX plate for a finished shot at the same length changes nothing for sound.
So the useful phrasing when something changes after lock is not "is picture still locked", it is "did anything move". Answering that question in the email saves a diagnostic pass on the other end. What the audio team then has to do is a conform, and how expensive it is depends almost entirely on how much work was already built against the old cut.
The Cut Decisions That Reach the Mix Stage
| What you do in the edit | What it costs downstream |
|---|---|
| Trim dialogue tight to the first and last syllable | No material to fade with, so the edit is audible and cannot be smoothed |
| Cut room tone out of the gaps | The dialogue editor has nothing to fill holes with and has to source or rebuild it |
| Nest or merge sequences | The AAF often delivers a stereo mixdown instead of the individual clips |
| Fix a bad edit with a crossfade | The mixer has to undo the fade to work on what is underneath it |
| Cut to a temp music track | The edit is shaped to a piece that is being replaced, and the replacement has to fit anyway |
| Delete alternate takes from the timeline | The cleanest version of a line may no longer be visible to anyone |
| Place sound design on whichever track was free | Content has to be identified and re-sorted before any of it can be routed |
Every row is a reasonable thing to do while cutting. The point of the table is not that any of them is a mistake, it is that each one has a cost that lands on someone else, and four of the seven are cheap to avoid once you know.
Air at the Heads and Tails Is Not Wasted Frames
A cut trimmed exactly to the first and last syllable looks clean on your timeline and gives the audio side nothing to work with. Every fade, every overlap, every attempt to hide an edit needs material either side of the cut point, and if the clip ends where the word ends then the material does not exist.
This is the same problem handles solve at export, but it is not the same fix. Handles pull extra audio from the source file beyond the cut, which recovers what the recording contains, and how many frames of it you get is an export setting worth agreeing in advance. What handles cannot recover is a performance that was never captured with air around it, or a take where the next line starts a quarter of a second after the previous one ends.
The practical version: when a dialogue edit is tight for pacing, that is a creative decision and it stands. When it is tight because you were trimming by habit, a few frames of air on each side costs nothing visible and gives the mix somewhere to work.
Room Tone Is Material, Not Silence
The stretches between lines are not empty. They carry the sound of the room, and matching that tone across a scene is a large part of what makes dialogue sound continuous rather than assembled.
When those gaps get cut out, or replaced with true digital silence, the dialogue editor has to rebuild the tone from whatever else is in the recording, and a scene where every pause is silent is a scene where every line sounds like it was recorded somewhere else. Leaving the underlying audio present through the gaps, even where you have muted or ducked it, is the cheapest single favour you can do the dialogue edit, and it changes nothing about how the cut plays for a client.
The same logic applies to a scene shot across two days or two locations. Where you know the tone changes mid scene, saying so is worth two lines in a note. Our reference on dialogue edit prep covers what that stage involves on the audio side.
Nested Sequences, Merged Clips and Multicam
These three are the most consequential items on the list, because they change what physically arrives rather than making someone's job harder.
Nested sequences. A nest is a sequence used as a clip inside another sequence, and on AAF export it frequently arrives as a single stereo mixdown of everything inside it rather than as the individual clips. In Production Expert's 2024 panel of audio professionals, checked August 2026, this is named as one of the recurring problems with Premiere exports specifically. What was six separable elements becomes one file, and no amount of skill on the audio side gets them apart again. Where a nest exists for organisational convenience rather than for a creative effect, flattening it before export is usually the fix. The export settings that matter are in our Premiere AAF export guide.
Merged clips. Audio and video merged into one clip is a picture side convenience that can carry the sync relationship into the AAF or lose it, depending on the application and the version. Worth testing once on your own setup rather than discovering it on a delivery.
Multicam. A multicam sequence has audio from every angle, and what the AAF exports depends on how the multicam is configured and what is active. The audio team usually wants all the angles' audio, not just the one you cut to, because the boom on the wide might be the usable take. That is a question worth asking rather than assuming either way.
The Temp Music Problem
Temp music makes a cut playable for a client and shapes the edit around a piece that is not going to be in the finished film.
Two costs follow. The obvious one is licensing, and it is not yours. The subtler one is that the composer or the music editor now has to write to a shape that was determined by somebody else's track, with the client already attached to the version they heard. That is a normal part of the job and nobody expects temp to disappear, but flagging where the edit is genuinely built on a temp cue, rather than merely accompanied by one, is useful information that costs one line.
Whether temp goes with the AAF is worth asking. Some teams want it as reference on its own track, clearly labelled. Others want it out, because a temp track that arrives unlabelled can end up in a stem.
Your Temp Mix Is Evidence
The rough mix you built for review is not throwaway work. Sent as the guide mix embedded in the reference video, it tells the audio team what you intended: where you leaned on a sound effect, where music carries a moment, where you ducked dialogue and by how much.
It is not going to be used as a mix, and nobody will treat it as one. It is the only record of your intent that does not require a meeting, and delivering it is a checkbox rather than a task. The full delivery specification, including where the guide mix sits and what format the reference should be, is in our guide to what a sound mixer needs before you send the AAF.
Two Vocabularies for the Same Timeline
Most friction in this handoff is not disagreement, it is two crafts using different words. The translation is short and worth having.
| You say | They say | Same thing |
|---|---|---|
| Locked sequence | Turnover | The delivery that starts audio work |
| Window dub, burn in | Reference, guide picture | The video with timecode in the image |
| Audio tracks A1 to A8 | DX, MX, FX | Dialogue, music and effects, sorted by content |
| Rough mix | Guide mix, temp mix | Your review version, sent as reference |
| Unlocking picture | Conform, reconform | Moving finished audio work to match a new cut |
| Deliverable | Print master, stems, M&E | What ships at the end, in its separable parts |
The last row is the one worth knowing, because it explains a request that otherwise sounds strange. When the audio team asks for dialogue kept separable from effects and music, they are not being fussy about organisation. An M&E deliverable is the full mix with the dialogue removed, which is what lets a distributor dub the film into another language, and it can only be produced if the three were never permanently combined. What each distribution route actually asks for at the end is set out in our deliverables checklist, and the sequence your locked cut kicks off is mapped in our audio post workflow guide.
What the Audio Side Automates, and What It Cannot
Some of what your cut creates gets handled mechanically now, and it is useful to know which parts, because it tells you what actually matters in the list above.
The sorting does. Tools like fPost read the audio content of each clip and classify it as dialogue, music or sound effects, then place it into the audio team's own session template, so a track called Audio 7 holding a mixture of content no longer has to be untangled by hand. Split stereo pairs from a Premiere export are detected and rebuilt automatically. Timecode and the import structure are preserved. That whole category, the untangling of how the material was organised, is largely solved.
What fPost cannot do, and no tool can, is recover material that is not there. A clip trimmed to the syllable has no air in it to find. Room tone that was cut out is not in the file. A nested sequence that exported as a stereo mixdown cannot be separated back into its parts. Alternate takes deleted from the timeline are invisible to everyone downstream.
That is the real division, and it is a more useful way to read this article than a list of dos and don'ts. Organisation is cheap to fix later. Material is not. Where the two conflict, protect the material.
Frequently Asked Questions
What counts as breaking picture lock?
Anything that moves frames. Reordering, trimming, extending or inserting shifts every audio event after it, so work already built has to be repositioned. Colour, titles, graphics and VFX shots delivered at the same length change nothing for sound. When something does change, saying whether anything moved is the most useful sentence in the email.
Should I leave room tone in the gaps between dialogue?
Yes. The audio under the pauses is what makes a scene sound continuous, and if it is cut out the dialogue editor has to rebuild it. Muting or ducking it in your cut is fine; deleting it is what causes the problem.
Do nested sequences cause problems in an AAF?
Often, yes. Nests frequently export as a single stereo mixdown of their contents rather than as separate clips, which is unrecoverable on the audio side. Where a nest exists for tidiness rather than for an effect, flatten it before exporting.
Should temp music go to the audio team?
Ask them. Some want it on a clearly labelled reference track so they can hear the intent, others want it out to avoid it ending up somewhere it should not. What is always worth flagging is where the edit is genuinely built around a temp cue, because the replacement has to work in the same shape.
Do I need to send alternate takes?
If you have them, yes. A line you rejected for performance is sometimes the only clean recording of that line, and the dialogue editor cannot ask for something they do not know exists.
Why does the audio team want dialogue kept separate from music and effects?
Because the final deliverables include stems and an M&E, the mix with dialogue removed, which is what a distributor needs to dub into another language. That can only be produced if the three were never permanently combined, which is why the split matters all the way back up the chain.
fPost prepares a turnover into an organised Pro Tools session, classifying content by what it contains rather than what it was named. Details at forte-ai.com/fpost.
.jpg)
