Transcription

After recording, open the Meeting Page panel and click Transcribe.

How it works

Audio is sent directly from your device to OpenAI’s transcription API (gpt-4o-transcribe) with speaker diarization enabled. The result includes:

• Speaker-labeled segments (Speaker 1, Speaker 2, etc.)
• Timestamps for each segment
• Full text transcript

The transcript is saved as .transcript.json alongside the audio file in your vault.

Privacy and data handling

Your meeting audio is handled with care:

• Direct to provider. Audio goes straight from your device to OpenAI’s API using your own API key. Pensum’s servers never receive, store, or process your audio.
• Local storage. The recording (.webm) and transcript (.json) stay as files in your vault. You control where they live and when they’re deleted.
• No intermediaries. There is no Pensum proxy or relay between your device and OpenAI for BYO users.
• Retention is yours. Configure audio retention in Settings (indefinite, 30 days, or 90 days). Delete recordings at any time.

For Pro All-in-One users, audio is routed through Pensum’s API proxy for billing purposes. Audio is not stored or retained beyond the transcription request. See our privacy policy for details.

Local transcription (advanced)

If you run a local Whisper-compatible server (such as Ollama, LocalAI, or whisper.cpp server), you can point Pensum at it instead of OpenAI. In Settings > AI features, set your Local transcription URL (e.g. http://localhost:8080). When configured, audio never leaves your device.

This is a power-user option. Pensum does not install or manage local transcription software.

Speaker identification

After transcription, click Identify Speakers. Pensum sends the first 5 minutes of transcript plus your attendees list to the AI, which maps speaker labels to names based on context clues (introductions, who addresses whom by name).

For this to work, add attendees to the meeting frontmatter:

attendees:
  - Jamie
  - Sam
  - Taylor

High and medium confidence mappings are applied automatically. Low confidence mappings are skipped.

Cost

Approximately $0.36 per hour of audio at standard quality, $0.18 at mini quality. You pay OpenAI directly via your API key.

Requirements

An OpenAI API key configured in Settings > AI features (separate from Anthropic).