Listen2Me / Windows x64
Choose where speech is processed and how words reach your text field. From game chat to a longer AI prompt, keep the controls close and the typing out of the way.
Download a local model once, then transcribe on your own computer without an inference server. Audio and recognition stay on your PC; no Hangry Labs account is needed. Initial downloads need internet. If you type the result into an online app, that app controls how it saves or shares your text.
Click your destination, hold Ctrl+Shift, speak, then release all keys to type. Change the shortcut in Settings for a comfortable one-handed gesture. Toggle mode starts and finishes with a press; hands-free recording lets you keep talking without holding keys. Keep the destination selected and review before sending.
Use Parakeet or Whisper for completed takes, or a supported Moonshine or Nemotron streaming model for live speech recognition. Run on CPU or a compatible NVIDIA GPU where supported. You can also connect your own compatible OpenAI-style ASR or Qwen streaming server. Languages, speed and accuracy depend on the selected model and hardware.
Map spoken phrases to text, Enter or Shift+Enter. Keep named presets for different games, projects or situations and switch between takes; only the selected preset is active. Add alternative wordings to the same rule. Keyboard actions are opt-in: choose distinctive phrases and test them before using them in chat.
With the default two-modifier shortcut, release one key and keep the other held for two seconds to discard a take. The shortcut works without reaching for the mouse. Cancellation stops further work, but cannot undo text already typed or recall audio already sent to a hosted server. Streaming may have sent audio before you finish.
Supported streaming models show a live preview and can optionally type as you speak. Use Toggle, hands-free mode or a single non-modifier shortcut for live typing. Holding modifier keys buffers typing until they are released. Recognition can revise earlier words; late revisions may stop insertion, leaving the final transcript available to copy.
If you selected the wrong field or typing was blocked, copy the final transcript from Your words instead of recording again. Up to ten completed takes remain in memory during the session. Clear or quit to remove the app's history; this is not a secure-erasure guarantee. Some games and protected text fields do not accept simulated typing.
Some games and text fields block automatic typing. Keep your text field selected and review the result before sending. Enter is sent automatically only if you enable a custom phrase for it.
This is a beta release. The installer is self-signed, so Windows may show a publisher warning. Get your copy from the official download above.
Listen
Download