gastrohem-media-processor

Audio transcription converts voice notes to text using insanely-fast-whisper, processing files in parallel and outputting structured JSON transcripts. Run 'Process media' to transcribe all audio from today's WhatsApp folder, or specify a date with '--scan-date' to target a particular day.

Claude Vision OCR extracts and summarizes image content into natural language Markdown files focused on relevant information. The Skill processes all images in your WhatsApp folders, generating `.md` summaries automatically without manual extraction.

Parallel processing with insanely-fast-whisper accelerates audio transcription across multiple files simultaneously, while Claude Vision handles image OCR. Process all media for a specific date or folder with a single command, skipping already-processed files automatically.

The Skill requires insanely-fast-whisper as a dependency and works with daily WhatsApp folders containing `.mp3`, `.ogg`, and image files. Default behavior processes today's date; use '--folder' or '--scan-date' flags to customize the input scope and location.

Command-line controls like '--scan-date' and '--folder' let you target specific dates (e.g., '24.10') or directories. The Skill defaults to today but supports flexible scheduling across all Gastrohem departments for date-specific or folder-specific requests.

Audio transcription generates per-file JSON transcripts for structured data integration, while image OCR produces Markdown summaries. Both formats are ready for review, further processing, or integration into downstream workflows.