Skip to main content

​
Start with clear speech

Use the best source recording available. Reduce background noise, clipping, echo, and music that masks voices. Avoid repeated lossy conversion when preparing an upload. Separate microphones and limited overlapping speech help Neolli distinguish speakers. Review speaker assignments in the source transcript before generating dubbing.

​
Improve transcription

Confirm the source language. Correct names, numbers, and specialist vocabulary in the source captions, then save those corrections before translating. For missing or inaccurate speech, listen to the original section. If the voice is hard to hear there, editing the transcript or using a cleaner source is more reliable than repeated transcription attempts.

​
Improve dubbing

Clear, representative speech helps voice cloning. Longer clean passages can improve the reference material, but a specific recording length is not a universal guarantee of quality. Listen for pronunciation, pacing, speaker consistency, and alignment with the video. Revise awkward translated phrasing, save the edits, and explicitly regenerate the affected dubbing segments. Use the full mixed audio to check music and background balance, and the dubbed-only audio to inspect speech. See AI Dubbing and Exporting.

​
Report a persistent problem

Send support the video or job ID, target language, and timestamps where the result is wrong. Describe what you expected and what you heard.