Your first recording

From launching the app to a searchable, exportable transcript.

1. Launch Zerds. On first run it downloads the default transcription model (~830 MB) with a progress indicator, then works offline.
2. Pick your audio. In Settings → Hardware, choose your microphone and the source for the other side of the call ("Others"). Windows loopback needs no separate permission. On macOS, system-audio capture needs Screen Recording permission; follow the app's permission prompt and relaunch after granting it.
3. Optional: turn on AI. In Settings → LLM, use the built-in local model or point Zerds at your own server (see AI summaries & chat).
4. Press Record. The Me and Others channels show live levels; if live transcription is on, text appears as people speak. You can mute or adjust the gain of either channel mid-call.
5. Press Stop. The session appears in the sidebar, named by date and time — searchable, replayable line-by-line, summarizable, and exportable to Markdown, SRT, or JSON.