Girish Konda

rostrum

Clean up talk recordings, by hand or through your agent.

Recording a talk in one take means restarting sentences, slides and clicks. Cutting those by hand takes hours, and one bad join ruins a clip. rostrum finds them, shows every proposed cut with its reason, and leaves the decision to you.

It grew out of the scripts I used to edit my Conf42 talk, which cut a single 60-minute take down to 33 minutes.

Source on GitHub Package on PyPI

  • Finds spoken cues such as “redo”, retakes, repeats, long pauses and dead air, and flags unclear speech for you to check.
  • Lets you play each join in a local review page, keep or cut it, and nudge its edges.
  • Renders frame-accurate joins with ffmpeg and checks sync at every join, on Windows, macOS and Linux. It also runs as an MCP server, with plugins for Claude Code and GitHub Copilot CLI and an extension for Gemini CLI.
Demo
rostrum on a short demo talk: init finds the cuts, the review page adjusts one, then import, render and verify.
A 58-second synthetic demo talk. The terminal shows the real output, trimmed to the main lines.

Try it

uv tool install rostrum
rostrum init talk.mp4

Needs Python 3.11 or newer and ffmpeg 6.1 or newer; pipx install rostrum works too. The first run downloads a speech model of about 480 MB. After that, everything runs on your machine. Full setup guide

Releases