wisprkey

wisprkey is a macOS voice productivity application designed to let users control their computer and enter text using their voice. Its primary workflow is simple: press a configured keyboard shortcut, speak naturally, and have clean, punctuated text appear in the active application.

The application provides system-wide dictation rather than limiting voice input to a specific editor or application. Users can configure shortcuts for dictation and other voice functions, allowing them to speak into documents, messages, notes, code editors, and other Mac applications. The company states that its local speech-to-text models support 31 languages with up to 98% accuracy.

wisprkey also includes local text-to-speech capabilities. Users can select text from documents, code reviews, messages, notes, or other applications and have it read aloud. The product currently advertises 64 voices and support for reading content in multiple languages.

The application is increasingly positioned as a broader AI computer assistant. Planned or emerging capabilities include an Agentic Cursor that can interact with the computer through voice, Conversation Mode for talking with connected AI agents, a Meeting Assistant for recording and querying meetings, and a Local Brain that learns the user's writing style and keeps contextual information locally on the Mac.

wisprkey is designed around local processing for its core voice functionality. Default transcription and built-in voices run locally on the user's Mac, and the company states that it does not collect local audio recordings, transcripts, or generated speech content through the default local experience. Optional cloud providers can be connected for additional AI, transcription, or voice functionality.