Open source, locally translate any video to English
SMRTR summary
mid-voice is a local, privacy-first video dubbing tool that replaces a video's audio with an English dub using the original speaker's cloned voice. It uses speaker-aware voice cloning without diarization, dynamic audio ducking, and pitch-preserving time-stretching to handle translation length differences — all running on your own machine with no API keys or cloud uploads required.
SMRTR provides this summary for quick context. The original article belongs to Hacker News.
Read the original article