A two-minute clip is easy to translate. A 90-minute lecture or a ten-lesson course is days of work for a translator, a narrator and an editor. Here AI does that work, and you check the script and pick the voices.
Upload a video from 1 minute to 2.5 hours and up to 6 GB: MP4, MOV, WEBM or MKV. The file goes up in parts, so if the connection drops, the upload picks up where it stopped. The picture is never re-encoded: the video track goes into the result as it is, only the sound changes.
ElevenLabs Scribe transcribes the speech and tells the speakers apart, then a language model reads the whole transcript before translating it, so names, terms and the way people address each other stay consistent from start to finish. You can fix character names and add a glossary, and the translation is redone with them. Every line is written to fit its time slot.
Then choose how it should sound. Voice-over is the documentary style: the original plays quieter underneath while a narrator speaks over it, with one narrator, a male and a female one matched to each speaker, or a voice per character. Dubbing replaces the speech entirely: Demucs separates the voices from the music and effects, and the new voices sit on the original soundtrack. A speaker's own voice can be copied from the video if that person has agreed to it, or you can pick a library voice. Subtitles skip the voices and give you the text in both languages.
Before paying for the whole video, you can render the opening minutes as a cheap preview. The script is paged, with search and a speaker filter: play the original and the translation of any line, fix a sentence and re-voice just that line. If the connection drops or the server restarts, the job carries on where it was, and nothing is charged twice.
The finished MP4 has two audio tracks, the translation and the original, plus subtitles in both languages, also available as SRT and VTT files. Translating a series or a course? Make it a project: the cast voices and the glossary are saved, so every new episode sounds like the last one.
Translate only what's yours: your videos, courses, interviews, podcasts, your YouTube channel, or documentaries you hold the rights to. You confirm this before you start.