AutoSubs v3.5.2
This update improves transcription accuracy for non-English languages and adds more control over how Whisper understands your audio.
✨ What's New
📝 Vocabulary & Context
You can now give Whisper models extra guidance before transcribing. Add names, brands, acronyms, technical terms, or a short note about the topic so AutoSubs is more likely to use the spelling and wording you expect.
This is especially useful for tutorials, product videos, interviews, podcasts, and videos with uncommon names or specialist vocabulary.
🛠️ Fixes & Improvements
🌏 Better Korean and Multi-Language Transcription
Fixed an issue where some Korean text could appear incomplete or corrupted in word-level subtitles.
This fix also improves handling for other languages that use multi-byte characters, such as Chinese, Japanese, Cyrillic-based languages, and emoji-containing text.
⏱️ More Natural Subtitle Timing
AutoSubs now better respects natural pauses detected during transcription, so short spoken phrases are less likely to be merged into the same subtitle.
💬 Better Punctuation Handling
Improved subtitle splitting around punctuation and closing quotes, so sentence endings like ." or ?) are handled more naturally.
Full Changelog: v3.5.1...v3.5.2