Transcribes audio files into text using a local whisper.cpp server with GPU acceleration.
This skill integrates local speech-to-text capabilities into the Claude environment by interfacing with a whisper.cpp server. It allows users to convert audio recordings, voice notes, and media files into high-accuracy text using the large-v3 model. Optimized for NVIDIA GPUs with CUDA support, it enables real-time transcription, shell-based recording, and Python integration, making it an essential tool for developers and researchers who require private, high-performance transcription without relying on external cloud APIs.
Key Features
01GPU/CUDA acceleration utilizing the large-v3 model
02Built-in troubleshooting for VRAM management and server connectivity
03Support for multiple audio formats including WAV and MP3
04Local speech-to-text transcription via whisper.cpp
05Seamless integration with shell commands and Python scripts
060 GitHub stars
Use Cases
01Transcribing recorded voice notes or meeting audio directly from the terminal
02Building automated speech-to-text pipelines in Python applications
03Generating text transcripts for local media files to assist in documentation
