Guide
It depends on the transcription mode and any cleanup step that runs afterward. Check both: audio can stay on your computer while a transcript is sent to a cloud model.
A downloaded model processes audio on your computer. Once the model is installed, transcription can work offline. Speed and accuracy depend on the model, your hardware, the microphone, and the recording.
The app sends audio to the provider you select, using your API key. That provider processes the recording under its own terms and data policies. Bringing your own key does not, by itself, prove which servers the app contacts.
The app sends audio to a service run by its vendor. The service may transcribe it directly or use another provider. Review the vendor's documentation for processing locations, retention, and subprocessors.
Transcription converts audio to text. AI cleanup may send that text to a different model to adjust grammar or formatting. If you need a local-only workflow, disable cloud cleanup as well as cloud transcription.
A price, subscription, or API-key requirement cannot answer these questions on its own. Consult the documentation and test the mode you actually use.
VoiceForge supports local Whisper transcription without a provider key, plus optional cloud transcription with your own key. Cloud AI cleanup is a separate choice and can send transcript text to the selected provider. Keep cloud transcription and cloud cleanup off for a local-only workflow.
Compare the setup choices and $79 one-time license →
Ten practical ways to plan, dictate, and review a useful first draft.
No other email, ever.