The local helper runs quietly in the background. It does not open a separate desktop window.
Find a recording and open it in the Editor. Your local episode folders stay on this computer.
Loading episodes…
Bring in per-person tracks from Riverside, Zoom, or anywhere — pick the folder, we do the rest.
Loading…
This session hasn't been processed. Open its folder and Process episode builds the analysis, transcript, auto-edit and show notes. Speech-to-text runs on this device with local Whisper, so recording audio is not uploaded. Episode files stay in the local folder; an optional Google Drive mirror can copy them directly from this browser. Show-note generation sends transcript text (not audio) to the configured LLM.
For large 4K masters, run the pipeline from the project root (replace the path with your session folder). Transcription uses verified local Whisper by default; the notes command sends transcript text only to the cloud LLM:
node tools/analyze.mjs "<session-folder>" --push node tools/transcribe.mjs "<session-folder>" --push node tools/auto-edl.mjs "<session-folder>" --push node tools/notes.mjs "<session-folder>" --push
Analysis is ready. Generate a first draft with the shared auto-edit engine, then refine it.
Render pops a console and runs the whole pipeline on this computer — everything lands in <folder>/out/:
episode.mp4 — the rendered videoepisode.srt + episode.vtt — captions / subtitlesepisode.mp3 — audio-only exportchapters.txt — YouTube chapter markers (also copyable from the Notes tab)If nothing opened: double-click setup-render-launcher.cmd in the project folder once, tick “Always allow” in Chrome’s prompt, then click again.
The buttons use a desktop launcher. If it is not set up, create the run script from the Process screen and run it in your session folder, or run the pipeline yourself:
node "<your tools folder>/episode.mjs" "<session-folder>" --push
Available when analysis, transcription, and the first cut are complete.
Open full editor ↗