v1.4.443

danielmiessler/Fabricv1.4.443Apr 9, 2026by github-actions[bot]

AI Summary

This release introduces optical character recognition capabilities for YouTube videos, allowing users to extract text from video frames.

Key Highlights

Implemented FFmpeg and Tesseract-based visual text extraction from videos.
Added configurable CLI flags for fine-grained control over extraction parameters.
Fixed multi-line `yt-dlp` outputs and resolved timestamp overflow bugs.

New Features

YouTube visual text extraction via FFmpeg and OCR

Full Release Notes

## Changes

### PR [#2073](https://github.com/danielmiessler/Fabric/pull/2073) by [sathvikc](https://github.com/sathvikc) and [ksylvan](https://github.com/ksylvan): feat(youtube): Implement visual text extraction via FFmpeg and OCR

- Implemented FFmpeg and Tesseract-based visual text extraction from YouTube videos, enabling OCR on video frames.
- Added configurable CLI flags for visual extraction parameters, giving users fine-grained control over the feature.
- Fixed support for multi-line `yt-dlp` outputs and updated syntax compatibility with modern FFmpeg versions.
- Refactored OCR processing to use bounded concurrency, context timeouts, and hardened CLI argument handling for improved stability and security.
- Resolved multiple reliability issues including Tesseract CLI argument handling, racy error handling, and timestamp overflow bugs.

### Direct commits

- Docs: make README badges clickable