LA Studio 0.2.0 introduces a complete local video dubbing workflow:
- Import audio or video.
- Separate vocals from background audio.
- Transcribe and translate timestamped dialogue.
- Generate replacement speech with local TTS models.
- Automatically rewrite overlong translations using an LLM and phoneme-based duration estimates.
- Fit generated speech to the original timeline.
- Mix the new dialogue with the preserved background audio.
- Review, edit, preview, and export the dubbed media.
The entire pipeline can run locally when local models and runtimes are selected, keeping media and inference data on the user’s machine.
Full Changelog: v0.1.10...v0.2.0