Version 1.2.1 — Gemini translation & first-run tour
Transcribe speech with Whisper, translate with on-device LLMs, Gemini, or GPT-5, polish in a built-in editor, and export — all running locally. Your files never leave your device.
Bring your favorite engine
From raw footage to finished subtitles — no command line, no juggling tools.
Powered by OpenAI's Whisper models through WhisperKit. Choose from seven model sizes — tiny through large-v3, including large-v3-turbo — to balance speed and accuracy. Auto-detect the language or set it yourself. Pre-download models from Settings with live progress, so the first transcription never keeps you waiting — and once a model is on disk, everything works completely offline.
Whisper isn't perfect every time. Click any line to edit its text right in place, open a popover to tweak start and end times, and jump the video to any segment with one click. Add, delete, merge, or split lines as you go. A built-in quality checker flags overlaps, fast reading speed, and overly long lines, with one-tap fixes. Full undo and redo throughout.
One app for the whole workflow — transcription, translation, editing, and export.
Seven Whisper model sizes (tiny to large-v3) run locally through WhisperKit, with automatic language detection. Works fully offline.
On-device LLMs (Qwen 3.5, Gemma 4, Phi-4), Google Gemini, OpenAI GPT (GPT-5, GPT-4.1, o3), Ollama, or any OpenAI-compatible endpoint — all in one unified model picker.
Export to SRT, WebVTT, and ASS/SSA, or burn subtitles into the video — with custom font, size, color, outline, and position.
Automatically flags overlapping timestamps, fast reading speed, and overly long lines — with one-tap auto-fix for most issues.
Drop in MP4, MOV, AVI, or MKV. MKV is remuxed on the fly with a bundled LGPL FFmpeg; Vorbis / Opus audio auto-converts to AAC.
A first-run tour walks you through the flow, translation models can be downloaded ahead of time in Settings, and the UI speaks English, Traditional Chinese, and Japanese.
Download SRTWhisper and turn your next video into subtitles in minutes.