In this guide you will learn how to generate an automatic transcript in Transkribus with a model for automated recognition. You can use a model you have trained yourself or one of the public models available in Transkribus.
First you will need to upload the document to Transkribus and let it be segmented into text regions and lines by the layout analysis.
Transkribus Expert Client
Running the Handwritten Text Recognition
To arrive at the recognition tool, open the “Tools”-tab in Transkribus, where you can find the “Text Recognition” section. Click “Run” and then “Select HTR-model”. Now choose the model you would like to use from this list and confirm with “OK”. Then select the pages you would like to generate a transcript of in the following window. Press “Run” to start the text recognition process and confirm the two appearing windows. This was already everything that needs to be done to start the automated transcription.
While your job is running you can check the progress of the recognition with the Jobs button, load other documents or close Transkribus. The job will proceed in the background. As soon as the process is finished, the program will ask you to reload. Please confirm.
You can now see the transcript in the text editor. If there is no text editor, you can open it with the transcription view.
By clicking on the baseline in the image, you will automatically arrive at the transcription line in the text editor. That also works the other way around: when clicking into one line in the text editor the program will jump to the corresponding line in the image.