Remotion AI Voiceover Skill — ElevenLabs TTS

How It Works

1. Define your script — Text for each scene in a config file
2. Generate audio — Script calls ElevenLabs API, writes MP3s to public/
3. Dynamic duration — calculateMetadata reads audio duration, sizes composition accordingly
4. Render — Remotion renders video with synced voiceover

Generating Audio with ElevenLabs

// generate-voiceover.ts
const response = await fetch("https://api.elevenlabs.io/v1/text-to-speech/{voice_id}", {
  method: "POST",
  headers: {
    "xi-api-key": process.env.ELEVENLABS_API_KEY,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    text: sceneText,
    model_id: "eleven_multilingual_v2",
  }),
});
// Write audio to public/voiceover-scene-1.mp3

Run: node --strip-types generate-voiceover.ts

Dynamic Composition Duration

export const calculateMetadata = async () => {
  const duration = await getAudioDurationInSeconds(staticFile("voiceover.mp3"));
  return { durationInFrames: Math.ceil(duration * 30) };
};

FAQ

Q: What TTS service does the Remotion voiceover skill use? A: ElevenLabs by default, but any TTS service that produces audio files can be substituted.

Q: Does the video duration auto-adjust to the voiceover? A: Yes. The skill uses Remotion's calculateMetadata to dynamically set composition duration based on the generated audio length.