Last updated: July 19, 2026
Duocap shows live translated captions for speech from your microphone and your computer's audio. Audio is processed in real time and never recorded to disk by Duocap.
While you are actively listening, audio segments containing speech are streamed over an encrypted connection to our speech-recognition provider (Soniox, processed in the United States) to produce captions, and caption text is sent to Google's Gemini API for translation. Audio is processed transiently for recognition and is not used to train models. Silence is not transmitted.
Transcripts, translations, and meeting summaries are stored only on your Mac (and in your own Apple Notes if you use the export feature). Duocap has no user accounts, no analytics, no trackers, and no server of its own that stores your content.
Listening only happens after you start it; a 3-hour automatic stop prevents forgotten sessions. You can end any session without saving, and delete saved history at any time from the app.
Questions: xu.tomi3@me.com