rip-video

This repository includes a TUI to download audio, split into parts, transcribe with ffmpeg-whisper, generate meeting minutes using a local Ollama model.

Install

Once published you will be able to install the binary directly:

cargo install rip-video

Until then you can still install from the repository:

cargo install --path .
# or
cargo install --git https://github.com/syzer/rip-video

Usage

1. Grab the media link (e.g. open DevTools Network tab and copy the request URL that contains videomanifest).
2. Launch the TUI with the URL:

   rip-video <MEDIA_URL>
   # optionally override the output audio path (default: audio.m4a)
   rip-video --output downloads/audio.m4a <MEDIA_URL>
   

3. The app downloads audio via yt-dlp, splits and transcribes it with ffmpeg-whisper, then generates Minutes and a Summary using local Ollama models.

If you prefer to rely on the clipboard, set the CLIPBOARD_TEXT environment variable before launching:

CLIPBOARD_TEXT="$(pbpaste)" rip-video

macOS convenience targets are kept in the justfile:

• just runs the command above.

• just create-minutes creates the local minutes model (one-time).

• just reset removes generated artifacts (audio, parts, transcript, minutes).

Prerequisites

• yt-dlp: brew install yt-dlp.
• just (task runner): brew install just or apt-get install just.
• ollama (required): install and run a local Ollama server. For macOS: brew install ollama and start it with ollama serve. Pull a model as needed (e.g., ollama pull llama3).

Ollama Setup

• Start the Ollama server:

  • ollama serve

• Pull the model used for Summary (required):

  • ollama pull deepseek-r1:14b

• Create the minutes model from the repo file (required for the Minutes tab):

  • just create-minutes
  • or ollama create minutes -f minutes.model

• Verify models:

  • echo "test" | ollama run minutes
  • echo "test" | ollama run deepseek-r1:14b