We Speak Input

macOS App

Speak instead of type, anywhere on your Mac.

Visit gtxistxgao.github.io/we-speak-input-releases

We Speak Input is a macOS voice-input tool that converts spoken words into text at the cursor location, working like a native input method across any application. Hold Fn to talk, release to type.

We Speak Input behaves like a system-level input method: press and hold the Fn key to record audio, release it to transcribe, and the resulting text appears automatically wherever the cursor is.

Transcription is powered by an embedded Whisper model (large-v3-turbo) that runs entirely on-device, so no audio is ever uploaded and the tool works without an internet connection.

Highlights

  • Free app for macOS 13 or later
  • Universal binary — Apple Silicon and Intel
  • Hold Fn to talk, release to type
  • Grants needed: microphone and accessibility

Key Features

  • Local, offline transcription

    An embedded Whisper large-v3-turbo model runs entirely offline — no internet required and no audio uploaded.

  • Works everywhere

    Because text is inserted at the cursor level, it works in any macOS application, just like a native input method.

  • Multilingual & bilingual UI

    Select multiple recognition languages from the menu bar; the interface toggles between Chinese and English or follows system settings.

  • Automatic punctuation

    Sentence-ending punctuation is added automatically so dictated text reads cleanly.