Deciphering grandma’s letters: Recognizing Old German writing on the Internet







Most people can only decipher Old German and Sütterlin script with great difficulty, if at all. But if you want to read letters from your ancestors, historical documents or the like, you can have such texts automatically transcribed on the Internet.

In principle, this is a normal letter and character recognition using OCR (Optical Character Recognition), just specialized in the conversion of the Kurrent, Sütterlin and Fraktur typefaces of earlier times. The usual OCR programs as well as online converters fail here completely.

In our experience, the Transkribus service, which emerged from an EU project, delivers good results. Via this portal you can upload the original document as an image in PNG or JPG format directly in the browser, have it analyzed and save the result in editable form as a PDF or Word document.

The text recognition is not perfect and needs corrections (left).  However, given the quality of the source material (right), the result is decent.

Enlarge

The text recognition is not perfect and needs corrections (left). However, given the quality of the source material (right), the result is decent.

You transcribe PDF documents by first taking a screenshot, saving it in one of the two image formats and then uploading it for analysis. In order to be able to use all functions, you must register for free. This registration also enables the download and use of the Transkribus software on the PC. The website provides you with detailed instructions on the service and the software.

Tip:

PDF24 at a glance: How to keep all PDF documents under control





Source link -63