Ogma

Recording meetings

Start, pause, resume — and why a crash never costs you the meeting.

Ogma is built for long (1–3 hour), in-person meetings recorded from a single microphone in the room.

Recording a meeting

  1. Hit record on the home screen. The app starts capturing from your configured microphone at 16 kHz mono — optimized for speech, small on disk.
  2. Pause / resume at any time (breaks, off-the-record moments). Paused time is simply not recorded.
  3. Hit stop when the meeting ends. Ogma finalizes the audio and kicks off the processing pipeline.

While recording, Ogma holds a wake lock so your machine won't go to sleep mid-meeting, even on battery.

Crash safety

Audio is not buffered in memory until you press stop — it is continuously written to disk as rotating 5-minute WAV segments. This is the core reliability guarantee:

  • If the app crashes, the OS blue-screens, or the battery dies, everything up to the last few minutes is already safely on disk.
  • On the next launch, Ogma detects the interrupted recording, repairs the WAV header of the segment that was mid-write, and recovers the meeting.
  • Worst case, a hard crash loses ~5 minutes — never the whole meeting.

Don't clean up audio files by hand

The segment files in the app data directory are the source of truth until a meeting is fully transcribed, summarized, and synced. Ogma treats source audio as sacred; you should too.

Practical tips

  • Place the laptop (or external mic) centrally — speaker attribution works off what the transcript "sounds like", so clear capture of every voice helps.
  • One recording per meeting works best. The pipeline is designed around a contiguous conversation.
  • Recording works with no API keys configured — you can capture now and process later once keys are set.

On this page