Overtone

Store each spoken word once, then let the player pick the line breaks

YSeen on Hacker News

If you publish one video both wide and vertical, this keeps a single transcript as the source and lets the player choose line breaks, instead of a subtitle file per crop.

Overtone treats a transcript as data rather than as a subtitle file. Each spoken word is kept once and line breaks are computed per picture shape, so a 16:9 cut and a 9:16 cut come from one source. There is a CLI that transcribes, re-cuts SRT or WebVTT, burns in and validates; a video.js plugin that serves whichever cut fits the player; and the .overtone.json format with exporters. Re-cutting needs no key and no network, by way of npx. Transcription wants ffmpeg and a provider key, or whisper.cpp locally. MIT, Node 22.12 up, built at Lont.

More like this

Rescript

Edit podcasts and video by editing the transcript, entirely on your device

Concat

A local video editor with offline captions and no watermark, desktop and phone

VisualFries

An MIT render engine your coding agent drives to cut raw footage into posts

CleanTake

Patch a ruined line of speech with the same moment from your other recorder

Narra

Voiceovers and burned-in captions made on your own device, with nothing uploaded

Concat

Open-source, local video editor built in Rust as a CapCut alternative