Voiceover (TTS) — AI setup — VidVertex Docs
Docs/AI setup/Voiceover (TTS)

Voiceover (TTS)

Settings → AI → Voiceover (text-to-speech). Providers: ElevenLabs and OpenAI TTS, same pattern — a key and a model per provider, plus a default voice.

  • Load voices fetches the account's real list (ElevenLabs includes your clones).
  • Test voiceover speaks a sample with the current settings.
  • Brands can carry their own voice (brand editor → Profile); the stack's hook/endcard tabs hold the volume slider.

Spoken lines are cached per provider, model, voice and text, so twelve variants sharing a hook line cost one call, and the editor's preview plays the same cached file the render uses.

TIP

ElevenLabs keys are scope-limited by default. Without the Voices: Read permission Load voices comes back empty while speaking still works — tick that permission when you create the key.