Local inference
Voixe uses local Core ML-capable speech models such as Parakeet and Whisper.
Privacy-first voice-to-text
On-device dictation for people who do not want cloud transcription.
Voixe runs transcription models on your Mac. Your voice is captured, transcribed locally, pasted, and then the workflow gets out of your way.
No account needed
There is no workspace to create before you can dictate into your own apps.
Simple permissions
Microphone captures voice, Input Monitoring listens for the hotkey, and Accessibility enables paste.
Privacy questions
What on-device dictation means in practice.
Does Voixe need the internet?
Voixe may use the network to download models and check for updates. The dictation flow itself is designed around local transcription.
Why not just use a cloud model?
Cloud models can be powerful, but many users prefer local speech processing for private notes, business messages, and day-to-day writing.
Is this only for technical users?
No. The intended workflow is normal Mac behavior: hold a hotkey, talk, release, and paste.