Superwhisper

Superwhisper is a dictation app built for people who want transcription without sending audio anywhere. It runs speech models locally on your device, so it works on a plane, in a secure environment, and without a subscription to a cloud service.

Key ideas

• On-device by default. Local models handle transcription with no network requirement; cloud models are available if you prefer them for a specific task.
• Modes. Each mode pairs a model with a prompt, so you can have one for email (polished, formatted), one for notes (verbatim), and one that translates as you speak. Switching modes changes what comes out of the same speech.
• Works everywhere. A global hotkey inserts text into any application.
• Custom vocabulary for names and technical terms.
• Voice-driven editing of existing text via selection plus instruction.

Who it suits

Journalists, lawyers, clinicians, and developers — anyone dictating material that should not transit a third-party server, and anyone who wants dictation that keeps working offline.

A free tier covers regular use; a paid licence unlocks the largest local models, all modes, and cloud options. One-time purchase and subscription options are both available.