A translation is mixed from three tracks: the dubbing, background sounds and the original voice. Their volume is adjusted on the "Sound" tab in the editor. Changes reach the file after "Re-voice", and no minutes are charged for that.
Tracks #
- Dubbing. The translation in the chosen voice. Cannot be turned off.
- Background sounds. Music, noise, room sounds separated from the speech. On by default via "Keep background sounds" at upload.
- Original voice. The muted source speech under the translation, like a voice-over. Off by default; enabled with the switch at upload or here.
Each track has a slider from −6 to +6, roughly 1 dB per step.
Dubbing volume #
The dubbing volume follows the voice in the source video: switching "Original / Translation" keeps the same level. Very quiet recordings get a gentle boost so the translation stays audible. If the dubbing seems quiet or loud, move the "Dubbing" slider and re-voice.
Noise and artifacts in the background #
Separating speech from music is not perfect. If the background track has leftover bits of voice or artifacts that were not in the original:
- Lower "Background sounds" by 2–3 steps or turn them off.
- If the music is needed, add it in a video editor on top of the translation downloaded without the background.
The background affects voice cloning too: the louder and busier it is, the worse the speech separates, see Why the voice does not sound like the original.