neofob / whisper-realtime

A front-end, "real-time" implementation for Whisper

Features:

• Keyboard emulation to "type" your transcription into the currently active text box (can be disabled via settings).
• Option to save audio (can be enabled via settings).
• Local logging of prior transcriptions.

---

While running, hold F9 to record.

Release when you're done talking.

This is a work in progress

It might crash.

It might crash your audio device (though, I think I've fixed that).

How to run

• Install Python 3.10.9

Then either

• git clone this repo

or

• Click the green Code button on the top right, Download ZIP, and extract to wherever you want it.

Then

• run.bat

todo

• Wrangle pyinstaller to correctly include whisper for a pre-built version.
• Allow custom hotkey for audio recording.
• Test on other systems to make sure it's capturing the audio device correctly.
• Clean up the jank.