Parakeet Audio Transcription

Acerca de

Leverage NVIDIA's Parakeet ASR model and Apple's MLX framework to perform high-speed, on-device speech-to-text processing for podcasts, interviews, and meetings. This skill enables exceptionally fast transcription—capable of processing an hour of audio in under a minute—without requiring cloud APIs or complex installations. By generating industry-standard SRT files locally, it ensures maximum privacy and efficiency for content creators, researchers, and professionals working on Mac hardware.

Características Principales

• Optimized for Apple Silicon MLX framework for high-speed local processing
• 0 GitHub stars
• Supports on-device execution for maximum privacy and offline use
• Compatible with common audio formats like MP3, WAV, and M4A
• Direct execution via uvx without requiring manual software installation
• Generates timestamped transcriptions in SRT subtitle format

Casos de Uso

• Generating searchable text and show notes from podcast episodes
• Creating accessible subtitles for video and audio content creators
• Transcribing professional interviews or meeting recordings for documentation