Special Collections and archives often hold unique and valuable material for which there is no published or online surrogate. Years ago, researchers would have to travel from library to library to access diaries, personal and professional papers, and histories that were not available anywhere else. One of the aims of library digitization is to make resources more readily available to researchers anywhere in the world.
After digitizing resources, libraries face another obstacle when it comes to making resources readily available to researchers, and that is the obstacle of discoverability. The excess of materials online makes finding any one resource online much more difficult. Libraries address this in a couple of ways. One way is to add rich metadata to digitized items so that online searches will locate material based on keyword and description text. Another way is to run Optical Character Recognition (OCR) software on scanned materials so that online searches can access the full text of the material.
OCR only works with printed text. That means all the handwritten letters, reports, and manuscripts in library collections are less searchable and therefore less discoverable via any online query. This is where Transkribus comes in.
Transkribus is a platform designed to learn and transcribe manuscripts, or handwritten pages, into computer readable text. This innovation allows Humanities scholars and Libraries working with handwritten materials to train the system on a handwriting sample, which will then be used to automatically transcribe the handwritten text into computer searchable text. Because Transkribus can be trained to recognize a specific handwriting sample, it can work with handwritten text in several languages.
Transkribus is available for personal download for volunteers who are interested in using the platform, scholars who are working with manuscript resources, and libraries who are digitizing handwritten material that they would like to make searchable. Each download helps with the further development and evolution of the platform by sending handwriting recognition data back to the Transkribus team.