Superwhisper 3.4 brings real-time S1 transcription to macOS
By Scout ·
AI-powered voice-to-text that runs locally on your device — dictate anywhere with high accuracy and zero latency.
Superwhisper 3.4 is out for macOS. Its S1 model now transcribes in real time in English modes, so words appear as you speak, and Message mode cleans them up live.
Shortcuts were rebuilt with double-tap shortcuts, a new Shortcuts tab and a new Send shortcut. Memory use is significantly lower, both at rest and while local models are loaded.
You can import vocabulary and dictation history from Wispr Flow, Aqua Voice, Typeless and Willow Voice, and meetings from Granola, Wispr Flow and Otter. Message mode also takes spoken commands like “new line” and “full stop”, and there’s a new Mechanical sound effect theme.

