flux-3-audio-dialogue — Claude Code skill: what it does, install
flux-3-audio-dialogue helps you direct FLUX 3 sound layer by layer: lines, voiceover, ambience, effects and music. Useful when a clip's audio comes out muddled.
MIT 181 installs ★ 138 updated 2026-09-09
npx skills add black-forest-labs/skills --skill flux-3-audio-dialogue -a claude-codeWhat flux-3-audio-dialogue does
This skill teaches you to direct sound in FLUX 3 prompts one layer at a time: speech, voiceover, ambience, effects, music and silence. Describe each separately, because one blurry description hands control of all of them to the model. A spoken line goes in quotes next to a visible speaker, or gets a voiceover or narration label so the model doesn't hunt for a mouth on screen. Add no on-screen text, no subtitles when text should stay out of the frame.
Scripts are sized to the clip's actual length: short sentences, one idea each, a pause around the payoff, odd names spelled phonetically. Effects are tied to an action, like a cup cracking on tile with one sharp snap. The mix says what leads and what sits underneath. A deliberately silent source clip uses generate_audio: false.
Repeating a voice description gets you a similar kind of voice, not the same performer, so a recurring speaker across clips won't hold. Exact wording, subtitles, final loudness and frame-accurate sync belong in post.
What it helps with
- attach a line to a visible speaker or mark it as narration
- fit a script to the clip's length
- keep café chatter from covering one important line
- produce a silent source clip
- debug a missed take by changing one thing at a time
Example prompts for the agent
- Use flux-3-audio-dialogue to write a FLUX 3 prompt where a weather presenter speaks one line to camera, no subtitles
- Shorten this voiceover line so it fits the clip, and spell the character's name phonetically
- Set up the audio layers for a café scene: her line in front, room noise low, piano under the last words
How to install flux-3-audio-dialogue
1Skills CLI (Vercel)
npx skills add black-forest-labs/skills --skill flux-3-audio-dialogue -a claude-code2Manually: put the files in place
~/.claude/skills/flux-3-audio-dialogue/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/flux-3-audio-dialogue/
1Skills CLI (Vercel)
npx skills add black-forest-labs/skills --skill flux-3-audio-dialogue -a codex2Manually: put the files in place
.agents/skills/flux-3-audio-dialogue/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add black-forest-labs/skills --skill flux-3-audio-dialogue -a cursor2Manually: put the files in place
.cursor/skills/flux-3-audio-dialogue/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install black-forest-labs/skills skills/flux-3-audio-dialogue2Manually: put the files in place
.github/skills/flux-3-audio-dialogue/SKILL.md1Skills CLI (Vercel)
npx skills add black-forest-labs/skills --skill flux-3-audio-dialogue -a gemini-cli2Platform CLI
gemini skills install https://github.com/black-forest-labs/skills --path skills/flux-3-audio-dialogueManually: put the files in place
.devin/skills/flux-3-audio-dialogue/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add black-forest-labs/skills --skill flux-3-audio-dialogue -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about flux-3-audio-dialogue
How do I install flux-3-audio-dialogue in Claude Code?
Run npx skills add black-forest-labs/skills --skill flux-3-audio-dialogue -a claude-code. The skill file sits at ~/.claude/skills/flux-3-audio-dialogue/SKILL.md.
What does flux-3-audio-dialogue do?
It guides how you write FLUX 3 prompts for speech, voiceover, ambience, effects, music and silence as separate layers, and how to troubleshoot a take that misses.
Will the skill keep the same voice across clips?
No. Reusing a voice description keeps the kind of voice, not the performer. Continuity across separately generated clips is something the skill assigns to post-production.
Source text
SKILL.md · MIT
--- name: flux-3-audio-dialogue description: Use when directing FLUX 3 audio, dialogue, or voiceover. Covers ambience, effects, music, silence, and timing. metadata: author: Black Forest Labs version: "1.0.0" tags: flux, flux-3, bfl, audio, dialogue, voiceover, sound-design --- # FLUX 3 Audio and Dialogue Name each layer separately: speech, voiceover, ambience, effects, music, or deliberate silence. One blurred description gives up control of all of them; every sound needs a physical source or narrative role. **Speech.** Quote the exact line, name the visible speaker (or label the line `voiceover`/`narration` so it is not searching for a mouth to belong to), and add `no on-screen text, no subtitles` when text is unwanted: ```text A weather presenter on camera in front of a stylized storm map, speaking directly to the lens: "Storm season is here, and this time, we're ready." Confident delivery, clean studio lighting. No on-screen text, no subtitles. ``` Voice anchors: age range, accent when relevant, register, energy, recording distance. Reusing the same direction preserves a kind of voice, not the same performer across generations. **Speakability.** Write for the clip's real duration: short sentences, one thought per line, room before and after the payoff; spell unusual names phonetically; shorten the line before speeding the delivery. A line that cannot finish comfortably needs a shorter script or a longer clip. **Effects are causal**, not a detached list: ```text As the cup hits the tile, it cracks with one sharp ceramic snap. ``` **Mix.** Say what leads and what stays under it. Keep background voices out when one line matters: ```text Her line is foreground and fully intelligible. Café chatter and espresso hiss remain low and diffuse. A restrained piano pulse enters beneath the final words without masking them. ``` **Silence and post.** Set `generate_audio: false` for a deliberately silent source clip. Reserve for deterministic post: final loudness, EQ, ducking, and fades; guaranteed wording or speaker identity; frame-accurate sync; subtitles and captions; continuity across separately generated clips. When a take misses, change one dimension at a time: speaker ownership, line length, delivery anchors, competing layers, action-to-effect causality, or generation versus post.
© Black Forest Labs · MIT · black-forest-labs/skills