Video Whisper — Local Video/Audio Transcription

To transcribe a YouTube video to text locally on a Mac, this Skill uses yt-dlp to extract audio and MLX Whisper to generate readable text entirely on-device. You just need to run the transcription script with the video URL to get the text output.

Yes, you can extract audio and transcribe podcasts without cloud services by running MLX Whisper locally. This Skill processes media entirely on Apple Silicon hardware, ensuring private transcription without uploading files.

Yes, you need yt-dlp and ffmpeg installed via Homebrew to transcribe Bilibili or Douyin videos. This Skill relies on yt-dlp to download audio from these platforms and ffmpeg for media processing before running the local transcription.

Yes, local video transcription with Whisper generates timestamps for segments. This Skill outputs both plain text and JSON files containing per-segment timings and detected language to help with indexing and downstream summarization.

The best way to get a searchable transcript from a local video file on Apple Silicon is using MLX Whisper. This Skill extracts audio from local media and runs on-device transcription for M1 through M4 Macs, producing text outputs directly in your local environment.

Local transcription requires a Python environment with mlx-whisper because it is the machine learning framework optimized for Apple Silicon. This Skill uses mlx-whisper to run the transcription model efficiently on-device without needing cloud APIs.