Documentation

What you need

• Windows (beta).
• Enough disk space for the app and its bundled model. No internet connection is needed to transcribe.

1. Add a recording

Open Verba and drop in a recording. Common audio and video files work — the audio is used, and each file becomes a named session you can rename anytime. You can queue several at once and let them run.

2. Transcribe & separate speakers

Verba transcribes the session on your device and automatically separates the speakers, pre-labeling them 상 (therapist) and 내 (client) by who speaks first. If the order's reversed, swap them in a click.

3. Edit

Fix a line inline, split or merge turns, and use undo/redo freely. Click any word to jump the audio to that moment. Verba keeps the original machine transcript, so edited text is always distinguishable from the raw output.

4. Korean supervision view

Switch to the annotated view for numbered speaker turns (상1, 내1, 상2…) in the format supervision reports use. Pauses longer than your set threshold appear inline as (침묵 N초) markers in every view, and clicking one jumps the audio to the start of the pause.

5. Export

Export to txt, csv, json, or xlsx, or save a self-contained .vrb bundle that carries the transcript and audio together — ideal for handing a session to a supervisor, who can open it and review it in context.

Privacy

Everything above happens on your machine. Your recordings and transcripts never leave your device.