1. Beginner’s Guide to Transkribus

Transkribus is an AI-powered platform designed to simplify transcription and digitisation of documents, from historical manuscripts to modern records.

Welcome to Transkribus! Transkribus is an AI-powered platform designed to simplify transcription and digitisation of documents, from historical manuscripts to modern records. With a wide range of recognition models for different document types, languages, and layouts, Transkribus helps users transcribe text accurately, offering both pre-trained and custom model options. 

This guide will walk you through the basics of using Transkribus for document transcription and introduce you to the different types of recognition available. By the end, you’ll be ready to start transcribing documents using the best tools for your specific needs.

1. Getting started with Transkribus

To begin, log in to the Transkribus platform and navigate to the Landing Page. Here, you’ll find a brief overview of your workspaces and documents, as well as Quick Text Recognition, which is ideal for most transcription tasks, especially if you’re just starting out.

What workspaces are there?

• Desk: The Desk is where the work happens. Here you have access to your Collections and Documents, upload or download them, start a text recognition and can edit transcriptions in the Transkribus Editor. 
• Models: In the Model Workspace you have access to an overview of pre-trained public models or custom private models. From here you can also start a custom model training.
• Sites: Transkribus Sites provides you with the possibility to create a searchable, online database of your documents which can be accessed by anyone, from anywhere. No coding or IT expertise required!

2. Using automatic text recognition in Transkribus

Here’s an overview of the main options:

• Quick Text recognition 

 What it means: If your document doesn’t require a specialised model, or if you’re working with commonly used languages and scripts, the Quick Text Recognition on your landing page is a good starting point.

How it works: Simply choose a language and drag &drop your page into Quick Text Recognition and get a result instantly

• Text Recognition using specific public models

 What it means: Transkribus offers a wide range of pre-trained models for various languages, time periods, and document types. You can also view and select a specific model to transcribe your text.

How it works: First, begin by uploading your document (PDF or image files) to your Transkribus collection. Select the document(s) you want to transcribe, and click the "Recognition" button. Then choose a fitting model (matching the language, material, etc.), and click “Start Recognition”.

Find a more detailed explanation in this Help Center article: Automatically transcribe documents

• Text Recognition using custom model 

What it means: If you regularly work with documents in a particular handwriting or style, consider training a custom model. Custom models are trained specifically on your documents, which can significantly improve accuracy.

How it works: Use correctly transcribed pages as training data to train a custom model. To gain an understanding of how to train a custom model, read through the articles of this section: Training Text Recognition Models

Good to Know: If you are working with documents with complex layout such as newspapers, forms, spreadsheets or registers, check out our trainable Field Models and Table Models. These help you to create more accurate transcriptions by not only extracting the text but also layout information in historical documents.

3. Review, correct and export

Once transcription is complete, go to the collection you have been working with to find the pages that were automatically transcribed. Click on the page to open the Document Editor and review the results or make any necessary corrections. Find out more about the Editor here: Document Editor 

Once you are happy with the result, you can either continue working on more material, publish the documents on Transkribus Sites, or export them in different formats.

For more help: Explore our Transkribus Help Center articles or contact our support team.