> ## Documentation Index
> Fetch the complete documentation index at: https://docs.gobihq.com/llms.txt
> Use this file to discover all available pages before exploring further.

# On-device and cloud transcription

> Two ways Gobi can process your audio, and what each means for your data.

Gobi can transcribe your captures in two ways. The difference is where the audio gets processed.

**Cloud.** Audio is sent to Gobi's own transcription service, processed there, and returned as text.

**On-device.** Audio is processed by a model running on your phone. The recording is not sent anywhere for transcription. On-device transcription keeps working in the background and while your phone is asleep.

Both paths run Gobi's own speech model. No third-party transcription service sees your audio on either one. The on-device version is the same model compressed to run on a phone; the cloud version runs it at full size.

Both produce the same kind of note. Pick based on what the conversation calls for: on-device when the audio shouldn't leave your phone, cloud when that constraint doesn't apply.