vox-jot-ocr-runtime

Vox Jot OCR Runtime

This repository hosts app-managed OCR runtime archives for Vox Jot. The runtime archive contains the ocr_runtime Python sidecar package and its Python dependencies for neural Screen OCR backends.

The archive does not contain Vox Jot credentials, signing material, user data, transcripts, or private app distribution secrets. It also does not contain OCR model weights; Vox Jot downloads model weights through the app's OCR model catalog.

Files

• ocr-runtime-macos-aarch64-all.tar.gz: macOS Apple Silicon runtime bundle.
• ocr-runtime-macos-aarch64-all.tar.gz.sha256: SHA-256 checksum consumed by the app before extraction.
• ocr-runtime-linux-x64-all.tar.gz: Linux x64 runtime bundle.
• ocr-runtime-linux-x64-all.tar.gz.sha256: SHA-256 checksum consumed by the app before extraction.
• ocr-runtime-windows-x64-all.tar.gz: Windows x64 runtime bundle.
• ocr-runtime-windows-x64-all.tar.gz.sha256: SHA-256 checksum consumed by the app before extraction.

Intel macOS is intentionally not published for the all profile. The neural OCR runtime depends on Torch >= 2.4, and compatible x86_64 macOS wheels are not available from PyPI for the standalone Python build target Vox Jot uses.

App Contract

Vox Jot downloads concrete files from this repository, verifies the checksum, extracts the archive under the app support runtime directory, and then marks neural OCR models runnable only after the runtime entrypoint, Python interpreter, and required modules are available.