What archival formats are available for export?

You can export transcriptions in ALTO XML, PAGE XML, and JSON formats, ensuring compatibility with most digital archival standards.

Does it support custom AI models?

Yes, advanced users can provide a custom YAML configuration to specify different models for segmentation and text recognition, such as using TrOCR models.

What languages does the htr-transcription skill support?

It supports Swedish, Norwegian, English, and Medieval scripts, with Swedish being the default language for transcription.

How should I handle multiple images to ensure efficiency?

You should batch all image URLs into a single call to htr_transcribe to minimize GPU pipeline overhead and significantly reduce processing time.

Can I use this skill for local image files?

Yes, but local files must first be uploaded using a file upload skill before their local paths or generated URLs are passed to the htr_transcribe tool.

Historical Document HTR Transcription

Name: Historical Document HTR Transcription
Author: AI-Riksarkivet

byAI-Riksarkivet

•

Data Science & ML

Transcribes handwritten historical documents into machine-readable text and archival formats using advanced HTR models.

The htr-transcription skill empowers Claude to digitize handwritten manuscripts and historical records by leveraging the Riksarkivet HTRflow MCP. It enables automated transcription of Swedish, Norwegian, English, and Medieval scripts with support for both single-page and spread layouts. Beyond simple text conversion, it generates interactive visual artifacts for verification and provides standard archival exports like ALTO and PAGE XML, making it an essential tool for researchers, archivists, and historians working with digitized archives.

Key Features

01Exportable data formats including ALTO XML, PAGE XML, and structured JSON for archival use.

028 GitHub stars

03Highly customizable pipelines via YAML configurations for specific model and segmentation needs.

04Batch processing support for efficient transcription of multiple document images in a single call.

05Generates interactive inline viewer artifacts with embedded dependencies for instant review.

06Multilingual handwritten text recognition for Swedish, Norwegian, English, and Medieval scripts.

Use Cases

01Batch transcribing collections of handwritten images via IIIF or direct public URLs.

02Converting scanned museum documents into searchable ALTO or PAGE XML formats.

03Digitizing historical letters, manuscripts, and archival records for academic research.

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add ai-riksarkivet/ra-mcp htr-transcription

For use in Claude.ai and ChatGPT

Download Skill