Handwritten Text Recognition (HTR)
AM Quartex enables transcription and full-text search of manuscript documents using Handwritten Text Recognition (HTR), an in-platform service customers can selectively apply. This proprietary neural network technology leverages artificial intelligence to recognize text in a multitude of scripts with remarkable accuracy, opening up manuscripts for detailed exploration by students and researchers.
Scroll down for more information about HTR, including samples - and be sure to check out the Manuscript Collections for more handwritten content.
Find answers to the most frequently asked questions about this feature and then click the sample record to try it out for yourself.
AM Quartex customers can use HTR to automatically generate transcriptions of handwritten material. Click here to find out more about Handwritten Text Recognition Transcription, languages supported and view samples of HTR transcription output.
What kind of searching does HTR enable?
This feature enables full-text search by keyword or string within handwritten documents using the in-document search box. Try searching 'nurse' in this document.
Does HTR work on other axes?
The technology is capable of reading text on any axis, including upside down. Search 'respected' on this seventeenth century letter.
Training the Technology
This unique service has been periodically enhanced using an established method devised by our partners which exposes the artificial intelligence to new scripts. By expanding the characters it can identify this training improves the accuracy of HTR's “linguistic assumptions” and recognize more hands.
Thanks to a corpus of documents from the Folger Shakespeare Library Quartex has enhanced the in-platform HTR functionality to read documents written in the script of 16th and 17th-century England known as Secretary Hand.