Every time I finished recording a video, the work wasn't done. The description, titles, chapters, captions — another 1-2 hours of typing stuff I'd just said out loud.
So I built postWRITER. Drop in a video, get back the transcript, description, chapter timestamps, title ideas, and social captions. There's a review page where the audio plays alongside the text — you fix a word, it saves, and that word goes into a personal dictionary. The next video comes out cleaner. The tenth comes out cleaner still.
How it works under the hood
Transcription runs on whisper-large-v3-turbo (via Groq — a 5-minute video transcribes in about 6 seconds). Writing is LLM-generated with a banned-words list to kill the AI voice ("delve", "in this video", etc.). Whole thing runs on a $5 VPS with FastAPI + SQLite.
What it doesn't do yet
English only (for now), links work for some platforms, and the first transcription of a heavy accent won't be perfect — the point is it converges as you correct.
It's free during beta. If you publish videos regularly, I'd genuinely like to know what still takes you the longest: https://postwriter.cn
Top comments (0)