dialogue-and-audio
Skill8 starsChanged 29 days agoMITView on GitHub
Dialogue, voiceover, subtitles, and mixing for AI video: TTS casting, per-line timing, SRT subtitles, loudness. Use when a video has spoken lines, narration, or needs its audio finished in post.
In this repository8 other files it gives its agents.
- assembly-and-deliverySkill
- character-consistencySkill
- cinematic-directionSkill
- model-adaptersSkill
- muse-google-veo-webSkill
- story-and-shotsSkill
- video-prompt-craftSkill
- video-qaSkill
The file
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 32
Time the lines: Every spoken line gets a start and end time in the shot-brief's beats.
- Distinctive: line 36
Write for speaking, not reading: Read every line out loud before generating.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Dialogue and Audio
AI video models are getting better at sound, but "better" still means unreliable: dialogue lands on the wrong beat, lips don't match, ambience is a guess. The professional default is to treat generated audio as a scratch track and finish sound in post. Plan for it from the start and you'll never be disappointed; depend on native audio and you will be, eventually.
The workflow
- Write the dialogue for the ear. Short lines, speakable words, timed to
beats (
references/voiceover.md). If a line can't be said in the seconds you gave it, rewrite the line, not the timing. - Cast the voices. One voice per speaker, locked for the project, same discipline as faces. Note pronunciation traps (names, invented words) before recording anything.
- Generate or record per line. TTS or human, one file per line, named systematically. Per-line files make retakes and re-timing trivial.
- Subtitle everything. SRT files, timed to the final cut
(
references/subtitles.md). Subtitles are accessibility, not decoration. Most viewers watch muted. - Mix it. Dialogue, music, effects balanced; loudness normalized for the
delivery platform (
references/mix.md).
Rules that don't bend
- Time the lines. Every spoken line gets a start and end time in the shot-brief's beats. Untimed dialogue is a wish.
- One voice per speaker, always. Switching TTS voices mid-project is the audio version of face drift.
- Write for speaking, not reading. Read every line out loud before generating. If you stumble, the TTS will too.
- Mute-test the cut. Watch the whole video with sound off. If the story doesn't survive, the visuals aren't carrying their weight.
- Keep the stems. Dialogue, music, effects as separate tracks until final export. Baked-in mixes can't be fixed.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

