Docs

Odin puts one layer over whatever is on your screen: live captions and translation from your computer's audio, an agent you can ask about what you're hearing and seeing, notes you can keep, and a learning mode for the words you keep meeting.

What you can do

How it works

On Windows, Odin captures system audio directly rather than a single browser tab, so anything your computer plays can be captioned. Audio is transcribed as you listen; translation happens on the same low-latency stream. Subtitles, agent answers, and learning prompts all render on one overlay that stays above your apps and lets clicks pass through to whatever is underneath.

When you're signed in, speech, translation, and agent requests run through Odin's hosted cloud, there are no API keys to configure. Developers can instead run with their own provider keys (ODIN_BYOK=true). Voice commands use the system dictation service, while media translation uses the Soniox real-time stream.

Platforms and requirements

Keyboard shortcuts

Plans and usage limits

Cloud capture time and Odin agent usage are metered per calendar month and reset on the 1st. Your live usage is always visible in the app under Account.

Basic

$5.99/month

10 hours of capture each month.

Odin agent usage included.

Enterprise

Quoted

Custom usage allowances per team.

Rollout planning and direct support.

Pro gives you 5× the usage of Basic: 50 hours of capture instead of 10, and five times the Odin agent usage, plus learning mode and priority support.

Learning mode is part of Pro. Local transcription and the features that don't reach a cloud provider aren't metered.

Billing

Privacy and data

Odin does not store your audio, transcripts, or translations on Odin servers. When you use cloud features, speech and text are processed by third-party providers and pass through. Optional usage telemetry (counts, not transcript text) can be turned off with ODIN_TELEMETRY=0. See the privacy policy for the full detail.

If something isn't working