Transcription Models
Understand the difference between local and cloud transcription in Vorn Voice.
Vorn Voice supports two ways to turn speech into text: local models that run on your device and cloud transcription that sends audio to a server.
Local models
Local models run entirely on your computer. Audio never leaves your device, dictation works offline, and there is no network latency.
Vorn Voice ships with models based on Moonshine and Whisper. They range from small and fast to large and accurate. See Choosing a model for details.
Cloud transcription
Cloud transcription sends your audio to a server (Groq) for processing. It supports more languages and can be more accurate for complex speech, but requires an internet connection.
There are two cloud modes:
- Vorn Cloud — Uses the Vorn proxy server. Included with your subscription at no extra cost.
- BYOK (Bring Your Own Key) — Use your own Groq API key for direct access.
See Cloud setup for configuration details.
Which should you use?
| Need | Recommendation |
|---|---|
| Fast dictation, English only | Local model (Vorn Standard) |
| Privacy-sensitive content | Local model (audio stays on device) |
| Offline use | Local model |
| Non-English languages | Cloud transcription |
| Maximum accuracy | Cloud transcription or Vorn Pro/Max local |