Dialogue and Sound in One Generation

Quick answer: A SceneMixer segment comes out with its sound already in it: each line spoken in that character’s voice, the speaker’s lips moving with the words, and the ambience of the place — rain, traffic, a clock. You do not record, dub or mix anything. You write the line with its voice tag, you write what the place sounds like, and the segment is generated with the picture and the sound as one thing. The download has the audio in it.
Still from “The Last Envelope”, an AI short drama generated end-to-end by SceneMixer (6 characters, 1 episode)
SampleThe Last EnvelopeDigital HD1 episodeProduced end-to-end by SceneMixer from the scriptWatch the episodes

Open a finished episode, play a segment with dialogue, and watch the speaker's mouth.

What comes out

Three layers, all generated with the picture. Voices: every tagged line, spoken in the voice chosen for that character, by the person whose lips move. Ambience: the sound of the place, from the Ambient line of the shot. Silence where you asked for it: a shot marked No dialogue has nobody talking. On top of those, the episode gets its own background score — chosen for the episode and mixed in under the voices, not generated with the clip — and the Music switch by the player turns it off. Background music in an AI short drama.

What you write to get it

A line: @Mara (low, steady): the words. (voice: 🔊Mara). The voice tag from the @ menu’s Voice tab is what says whose voice speaks; without it the line may be spoken in some other voice, and the editor warns you after saving. Dialogue lines.

Ambience: at the end of a shot, Ambient: rain on the glass, a clock in the hall. Say what is audible, in a few words. Nobody speaking: No dialogue.

Lips: the person speaking moves their mouth; give anyone else in the shot lips closed, no lip movement so they do not mouth the line.

A storyboard segment: location, time and weather fields, then numbered shots with durations, asset chips and spoken lines
Lines with their voice tags, and an Ambient line at the end of a shot. That is the whole soundtrack instruction.

Where the voice comes from

Each character has one voice, chosen in step 2 — matched to the character, previewed, changeable, or uploaded by you. Every line tagged with that character is spoken in it, in every episode. To change how a character sounds, change the voice on the character, then regenerate the segments you want updated. How character voices work.

Which language

Lines are spoken in the project’s dialogue language, set when the project was created. Write the lines in that language. A German project speaks German; the interface you use to write it does not matter. Choosing a dialogue language.

Changing the sound afterwards

SceneMixer has no mixer and no separate track to edit; the sound is part of the clip. You can still change it without regenerating: Redub, under the player of a finished segment, changes the lines or a character’s voice, and Trim cuts a stray word at the start or end. Or change the words or the delivery note and regenerate the segment. If the voice is wrong, check the tag first, then the voice on the character. Change a line after the video is made · Change a character’s voice.

When the sound is off

Wrong voice on a line: the tag names the wrong character, or is missing. Listener moving their mouth: add lips closed, no lip movement. Line rushed or cut: the shot is too short for the words — about three words a second in English. Shot seconds. Too quiet a room: the shot has no Ambient line; add one. Then regenerate that segment; the old take stays in Version History.

You wantWrite
A line in her voice@Mara (how): the words. (voice: 🔊Mara)
Rain and a clock under the sceneAmbient: rain on the glass, a clock in the hall
Nobody talkingNo dialogue
The other person not mouthing the line@Jules lips closed, no lip movement
A voice from off screen(voice-over) at the start of the delivery note

FAQ

Do I need to add voices afterwards?

No. The lines are spoken in the generated clip, in the character's chosen voice, and the lips move with them.

Can I change a line after the clip is made?

Yes. Use Redub under the player; it lists the segment's lines for you to edit.

Can I use my own voice?

Yes, on a paid SceneMixer plan. Upload a voice on the character in step 2; every tagged line for that character is then spoken in it.

Is there background music?

Yes. SceneMixer gives every newly made episode its own score, chosen for the episode and mixed in under the voices; episodes that had clips before scoring rolled out, in early October 2026, are not scored. The Music switch by the player turns it off — the switch shows only on clips that have a scored version — and downloads follow the switch.

The wrong character spoke the line.

Check the voice tag after the line: it names the speaker. Fix it and regenerate the segment.

Does the downloaded file include the sound?

Yes. Voices and ambience are in the clip, and in the composited episode, with the episode’s score unless you turned the Music switch off.

Hear your first line

Write one line with its voice tag, generate the segment, and listen.

Open the editor

Updated November 6, 2026