AI voiceover

AI voiceover in 31 languages, directed scene by scene

Scenes AI narrates your video with one of three speech engines — Gemini 3.1 Flash TTS, or either ElevenLabs rung, which offer 68 voices each — and you can set emotion and speaking pace per scene rather than for the whole video. Signed-in teams can also clone their own voice and narrate with that. Alternatively the video model can generate its own lip-synced dialogue, which is what you want for drama.

Three ways to handle audio

Which one is right depends on whether people are speaking on screen.

  • Separate narration: the video is muted and a speech engine narrates over it. Best for explainers, tutorials and faceless content.
  • Embedded dialogue: the video model generates the audio itself, lip-synced to the character. Best for drama and anything with people talking on camera.
  • No audio: silent video, for when you will add sound elsewhere.

Direction, per scene

Narration is not one flat read from beginning to end.

  • Nine emotions, set per scene rather than per video.
  • Three speaking rates — slow, medium and fast — also per scene.
  • One engine takes a natural-language style instruction, so the emotion genuinely changes the performance rather than just the speed.
  • Scene durations are reconciled against the real length of the generated audio, so narration is not cut off mid-sentence.

Picking a voice

Voices are filtered by the language and, if you want, gender you asked for, and a short list is presented for you to choose from rather than making you audition thirty of them. Characters also carry a fixed voice description, so a character keeps the same voice across scenes and episodes.

Working in other languages

The language you choose governs everything the audience hears and reads.

  • The setup form offers 31 languages, and the speech engines take the language from the script itself.
  • Captions are produced in the same language as the narration.
  • Video engines are told the language explicitly, so quoted dialogue is not silently translated into English.

FAQ

AI voiceover questions

Can't find what you're looking for? Visit help and support or email hello@usescenes.com and a human will reply.

The setup form's picker offers 13 narrator voices on Gemini 3.1 Flash TTS, each with a sample you can play; the two ElevenLabs engines have 68 each. They are filtered by your language and, optionally, gender.

Your first video starts with one sentence.

Describe what you want. The director writes the script, casts it, shoots it, adds voiceover and renders the final cut. Every spend shows its price first — approve each one, or let Auto mode run it.

Pay as you go · No subscription · Credits never expire

Episode one of “Last Call at the Starlight”, as rendered