The same voice in a language it never recorded.
23
languages from one cloned reference
Dubbing usually means casting a stranger. Cloning inverts that: the identity is a request parameter, so the same speaker says the translated line.
Hear the identity survive the translation
Three takes below: one cloned voice, one sentence, three languages. Nothing was re-recorded and no second speaker was cast. The reference never spoke Spanish or German.
audio
the three takes on their own, if you would rather scrub them
generated 2026-08-05
English
“We shot the whole thing in a single afternoon, with one camera, in a room we had borrowed for four hours. Nobody expected it to work, not the crew, not the client, and honestly not me either. We had no second location and no time to reshoot anything, so every take had to be the one we used. And then it did work, which is still the part I find hardest to explain.”
Mia · 19.0 s · video dubbing · source language
Spanish
“Lo grabamos todo en una sola tarde, con una cámara, en una sala que habíamos pedido prestada durante cuatro horas. Nadie esperaba que funcionara: ni el equipo, ni el cliente, ni yo, si soy sincera. No teníamos una segunda localización ni tiempo para repetir nada, así que cada toma tenía que ser la definitiva. Y entonces funcionó, que sigue siendo la parte que más me cuesta explicar.”
Mia · 21.9 s · video dubbing · same voice, translated
German
“Wir haben alles an einem einzigen Nachmittag gedreht, mit einer Kamera, in einem Raum, den wir uns für vier Stunden geliehen hatten. Niemand hat erwartet, dass es funktioniert, weder das Team noch der Kunde, und ehrlich gesagt ich auch nicht. Wir hatten keinen zweiten Drehort und keine Zeit für Wiederholungen, also musste jede Aufnahme sitzen. Und dann hat es funktioniert, und das ist bis heute der Teil, den ich am schwersten erklären kann.”
Mia · 22.8 s · video dubbing · same voice, translated
What this is, and what it is not
Gandr synthesises the dubbed audio and carries the voice across languages. It does not translate your script, split a voice out of a mix, or mux audio back onto video. Those stay in your pipeline.
- One reference clip covers the whole catalogue.
- Per-request prosody and pacing, so a translated line can be delivered at the pacing the scene needs rather than the pacing the words happen to have.
- Nothing is metered, so localising the back catalogue is not a per-title decision.
Notes
Does the cloned voice really hold up in another language?
Check it yourself. The three clips above are one cloned reference speaking English, Spanish and German, rendered on the production API. Dated specimens for eight languages live at /languages/.
Do you translate the script?
No. Send us the translated text and a reference clip, and you get the dubbed audio back.
A key and one stream to build this on, the same production API this page measures.
Get a key for this use case