WhisperDesktop
Windows desktop app that uses GPU acceleration to transcribe audio files and live mic input with OpenAI Whisper.
About
WhisperDesktop is a Windows application that brings OpenAI's Whisper speech recognition model to the desktop with GPU-accelerated processing via DirectCompute, making it significantly faster than the standard Python implementation without requiring heavy dependencies. It supports most common audio and video formats through Windows Media Foundation and includes a live capture mode with voice activity detection so you can transcribe in real time from a microphone. The application also exposes a C++ API and a C# wrapper for developers who want to integrate speech recognition into their own Windows software. WhisperDesktop is free and open source and requires a 64-bit Windows machine with a Direct3D 11-capable graphics card.