Turn a tag's videos into text
- Export by tag. Exported files carry each post's tags, and
export --tag <name>exports only one tag's posts.pushno longer puts back a tag that was removed on another device after the export. - Transcribe by tag.
transcribe --tag <name>(or several archive ids) skips posts without audio or video and ones already transcribed. Add--runto transcribe them one at a time on this computer. - Instagram, Facebook and X videos.
--runnow transcribes them by downloading the copy saved in your archive, so yt-dlp is only needed for YouTube and TikTok links. - No separate Whisper on a Mac. With the desktop app installed, the CLI uses the app's WhisperKit and the models you downloaded there, including a CLI installed with Homebrew or npm.
- On-screen text. The new
screen-textcommand reads the captions and text shown in videos and saves them as a note thatexportincludes (macOS, desktop app 0.8.19 or later).
Also in this version
- executor also answers Ask AI questions asked on a reader selection on your phone.
- whisper.cpp finds a model picked by its WhisperKit name, and falls back to WhisperKit when the model is missing.
Note
The desktop app 0.8.19 bundles this version. Its WhisperKit helper reads MP4 and MOV video without ffmpeg; with an older desktop app, transcribing a video still needs ffmpeg and screen-text is unavailable.
Signed and notarized macOS arm64/Intel builds plus a verified Ubuntu 22.04+ Linux x64 build. The checksum-verifying native installer and keyless Sigstore provenance bundle are included.
Install / update
brew upgrade social-archiver-clinpm install -g social-archiver@0.1.23curl -fsSL https://github.com/hyungyunlim/obsidian-social-archiver-releases/releases/download/cli-v0.1.23/install.sh | shWindows users can update from the Microsoft Store.