Introducing the first major Transkribus release of 2025: What's changed and why?

Welcome to 2025 with Transkribus. This first major release of the year brings significant updates aimed at improving how you work with documents. Rather than just listing features, we want to explain the reasoning behind these changes and how they support your workflows. Let’s explore what’s new and why it matters.

Redesigned action bar: improving consistency and control

We redesigned the action bar to ensure a consistent experience across the entire platform. The previous version had limited filtering and sorting options, which made navigating large collections cumbersome. Now, you can:

• Filter and sort documents by metadata such as creation date or name
• Switch between list and card views for a customised workspace
• Enable or disable columns to maximise screen space

These changes were made to reduce friction in everyday tasks and ensure the interface feels familiar, no matter where you are on the platform.

New batch actions: reducing repetitive tasks

Users working with large collections needed better bulk management tools. The new batch actions let you:

• Select multiple documents with checkboxes
• Apply bulk actions, such as running recognitions or managing labels

By cutting down on repetitive actions, this update helps you complete large tasks more efficiently, saving time and effort.

Document labels: building flexible workflows

Labelling documents is key to managing large projects. With this release, we are introducing a new system that lets you assign labels to documents. The new label system allows you to:

• Assign predefined labels (custom labels might follow in the future)
• Filter documents by labels to streamline workflows

This approach lets you tailor your organisation system based on the needs of your project, allowing you to create an advanced filtering system. This will help you label documents according to certain workflow steps, training data status, or material conditions. 

Recognition redesign: simplifying a critical process

The recognition process is central to Transkribus, but it previously required you to go through a cumbersome process with a very loaded user interface. The redesigned interface now:

• Keeps recognition within context using a slide-over panel
• Shows real-time credit usage and balance
  
• Offers quick access to favourite models
• Allows you to select either personal or collection credits

These adjustments make starting recognitions quicker and clearer, focusing on what matters most — getting accurate results.

New upload system: robust handling for large files

Uploading files with Transkribus was a pain point, especially for large collections. Our new and improved upload offers the following improvements:

• Drag-and-drop functionality for files and folders—you can drag across multiple folders containing multiple images
• Automatic document creation per folder
• Real-time progress tracking with improved reliability
• Easily upload several thousand images at once (depending on your internet speed)

The main goal here was to ensure uploads are robust, letting you upload multiple folders and larger amounts of files at once.

Quick navigation and search: faster access, less frustration

We found that searching across collections and documents needed to be faster and more intuitive. Now you can:

• Instantly search collections, documents, file names, and full text from the same search bar
  
• Use a dedicated full-text search page for deeper queries
• Experience a clearer and more responsive interface
• Search only in the current location (document or collection)

These updates reduce the time spent searching, helping you stay focused on analysis and transcription. More search categories will be introduced in the future.

Redesigned models space: making model selection easier

Accessing and testing recognition models was previously not straightforward. We have, therefore, redesigned the entire models space. The improved space now:

• Offers improved filtering and sorting
• Provides direct testing of models in the model details using Quick Text Recognition
• Shows detailed model information without leaving the page
• Gives you more control over your models and the information associated with them

This streamlines the process of selecting the right model for your documents, especially for specialised use cases, and also lets you manage your own models better. 

Training lab: centralising model training management

Training your own models is an important capability. The new training lab:

• Centralises management of all model trainings
• Allows direct access to all models trained in one central place

By bringing all training processes together, this feature simplifies custom model creation, supporting advanced recognition needs. Expect more improvements to this space soon.

Editor improvements: focusing on readability and clarity

Complex layouts and overlapping tags made editing difficult. The updated editor now:

• Displays structural tags outside the text region
• Improves readability by clearly showing text pane sections
  
• Lets you enable or disable the layout elements with one click

These improvements help reduce confusion and streamline the editing process, especially for complex historical documents.

Improved version history: tracking changes with confidence

Saving your work and tracking document changes needed more transparency and stability. The enhanced version history:

• Allows side-by-side comparisons with highlighted differences
• Tracks safe and unsafe changes before saving
• Introduces local autosave to prevent data loss during connection issues

These enhancements give you more control, ensuring no critical edits are lost or overlooked.

New layout reordering in the editor: correcting recognition order

Recognition jobs sometimes output incorrect layout orders, which made editing more complicated. The new layout reordering feature allows you to:

• Automatically reorder regions or baselines of a page
• Quickly correct layout issues without manual adjustments

This feature simplifies working with complex documents, ensuring that the structure matches the intended reading order.

New collection settings: centralising management

Managing collection settings used to be fragmented. The new collection settings section now lets you:

• Manage general settings of a collection
• Add or remove collection members with ease
• Organise textual, structural, and relational tags of a collection in one central place

These improvements centralise control and make it simpler to keep collections organised and accessible to the right users.

New streamlined invitation feature: simplifying collaboration

Collaboration is at the heart of many projects. To make it easier, we have introduced a streamlined invitation feature that allows you to:

• Invite users who do not yet have a Transkribus account
  
• Generate an invite link for users to join a collection directly
  
• Invite multiple users at once
• And invited users can accept invitations with just one click in their email
  

This feature was designed to reduce barriers when working with new collaborators, ensuring smooth onboarding and faster project starts.

What’s next for 2025

This release was driven by feedback and real-world use cases and represents a major step forward. Each change was made to support efficient workflows, reduce complexity, and make powerful tools more accessible.

Looking ahead, we remain committed to refining Transkribus based on your needs. Stay tuned for further updates and thank you for being part of the Transkribus community—let's unlock every doc.