mehbylabsmehbylabs/creator-skills/skills/creator-audio/SKILL.md

creator-audio

Skill7 starsChanged 16 days agoMITView on GitHub

Make voiceovers, background music, and sound effects with whatever the user has — ElevenLabs, Google Gemini API (Gemini TTS, Lyria), Google Cloud TTS, Azure Speech, OpenAI, Higgsfield CLI, or Google AI Studio via computer use — with automatic fallback and dialect-aware routing for Arabic (Fusha, Egyptian, Gulf, Saudi, Levantine, Iraqi, Maghrebi). Use when the user asks for a voiceover, narration, text-to-speech, Arabic or dialect voiceover (تعليق صوتي، صوت بالعامية), reading a script aloud, a podcast intro, background music, a jingle, a sound effect, ambience, or foley. Not for dubbing or voice cloning.

In this repository10 other files it gives its agents.
- AGENTS.mdAGENTS.md
- creator-adSkill
- creator-brandkitSkill
- creator-characterSkill
- creator-explainerSkill
- creator-imageSkill
- creator-product-photoSkill
- creator-setupSkill
- creator-thumbnailSkill
- creator-videoSkill

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:
Never generate a voice that imitates a real, identifiable person without their consent.
line 78
Distinctive:
Don't mix voices from different providers in one piece.
line 79

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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.

Keep what fits and drop the rest. Your version, ready for your agents.

Files

SKILL.md

Creator Audio

Pick the mode from the request, then follow references/audio.md for prompt and voice guidance.

Voiceover

Language — detect it from the request/script. For Arabic, ask which dialect if not clear (Fusha, Egyptian, Saudi, Emirati/Gulf, Levantine, Iraqi, Moroccan…) and map it to --lang with references/arabic-voice.md.
Script — if the user gave a topic, draft it (≈150 words/min; Arabic ≈ 120–130). For a dialect, write it in that dialect's spelling. Show it; wait for approval.
Split into paragraphs/scenes; save as vo/01.txt, vo/02.txt, … Arabic: add tashkeel to every word whose meaning depends on its vowels (fractions, دَين/دِين، عَقد/عِقد، ذَهَب…, passive verbs, numbers in words) before any TTS, then run python3 scripts/creator.py tashkeel-check --text-file vo/01.txt until it says ok (references/arabic-voice.md).
Generate the first file and have the user listen before doing the rest:
```
python3 scripts/creator.py speech --text-file vo/01.txt --out vo/01.mp3 --verify [--lang ar-EG] [--voice <id>] [--style "warm, confident"]
```
--verify transcribes the result and lists words read differently from the script (verify.differences, numbers first). Re-voice any line whose number, fraction, name or negation changed.
Pin the provider and voice from the result (--provider <id> --voice <voice>, keep --lang) for every remaining file so the narrator never changes.
python3 scripts/media.py probe vo/01.mp3 to confirm duration; report total length.

needs_browser with aistudio-browser → follow references/browser-aistudio-browser.md.

Music

python3 scripts/creator.py music --prompt "<genre, instruments, bpm, mood, arc>" --duration 60 [--instrumental] --out music/bed.mp3

Beds under speech → always --instrumental. The router rejects silent or truncated takes and extends short ones to the exact --duration (see notes); still listen once.

Sound effects

python3 scripts/creator.py sfx --prompt "<source, action, space>" [--duration 2] [--loop] --out sfx/<name>.mp3

Results

`status`	Do
`ok`	Report file, length, provider/voice
`blocked`	Explain; offer a revised script/prompt
`needs_tashkeel` (exit 5)	Nothing was sent. Add tashkeel to each listed word for the intended reading, rerun
`ok` + `verify.status: check`	Listen to each listed difference; fix tashkeel/wording and re-voice lines with a real misreading
`needs_browser`	Drive the provider in `next` (computer use is on by default)
`failed`	Name the cheapest provider to add and its `connect` command from the JSON (user runs it in their terminal; `references/connect.md`)

Rules

Never generate a voice that imitates a real, identifiable person without their consent.
Don't mix voices from different providers in one piece; regenerate instead.
Output format follows the extension (.mp3 default, .wav for editing).

References

references/audio.md — voices, scripting for the ear, music and sfx prompts
references/arabic-voice.md — Arabic dialects: script rules, --lang, routes, voices
references/browser-aistudio-browser.md — AI Studio speech playground (manual route)
references/providers.md — speech/music/sfx order and fallback
references/media-ops.md — combining audio with visuals

Download now (ZIP)

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

What did you use it for? e.g. Reviewing pull requests in a Rails app

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.

Check your setup, free