voiceover
voiceover
Run every graph with the bundled client: python scripts/comfy_client.py --url $COMFY_URL run workflows/<file> --set NODE.INPUT=VALUE --out out/. Upload input files first with python scripts/comfy_client.py upload <file>. Before the first run on a server, check it with python scripts/preflight.py --url $COMFY_URL workflows/<file>. Node packs, model files, VRAM and weight licenses for this skill are in references/requirements.md. Use open-weights models only, and never swap in a hosted or closed model.
Graph
voice_chatterbox_multilingual.api.json: set 1.text and 1.language (for example English (en)), and 1.seed. Node 2 LoadAudio is the voice reference. Upload a clean 5-15 s clip and set 2.audio. With no reference, delete node 2 and the audio_prompt link in node 1 (the rosesfield README in references/ describes this). If you need the 40 s trim, it is in node 4.
Voice conversion
voice_change_chatterbox.api.json (Chatterbox VC) re-voices recorded speech: set 3.audio to the source speech, 2.audio to the target voice and 1.seed. Only use a target voice with the speaker's consent.
Narration rules
Condensed from Fuobe's narration skill; the full copy is in references/fuobe-narration-rules.md.
- Use one consistent narrator voice, reference clip and seed for the whole film. Log them in the story bible.
- Deliver speech only. Music and SFX go to
music-sfxand are mixed later infilm-assembly. - Write for the ear: short sentences, numbers spelled the way they are spoken, no markup. Render one line per shot so lines can be re-timed.
- About 2.5 words per second. If a line runs longer than its shot, cut words; don't speed up the audio.
- Only clone a voice with the speaker's consent.