Guess who is familiar with a thousand different writers, speaks German, Lower German and Latin and has a favour for 17th century manuscripts? That is the “Acta_17”-model created by Dirk Alvermann, Elisabeth Heigl and Anna Brandt of the University of Greifswald. Thanks to them the model is publicly available in Transkribus – feel free to try it out if it fits to your material! The training data is based on legal texts and court writings from the Responsa of the Greifswald Law Faculty and can cope with simple German and Latin abbreviations. There is an HTR+ as well as a PyLaia-version of the model and in both cases nearly 600 000 words have been trained. The Character Error Rate is about 6%.
In this regard we would like to thank all users, who have shared their models with the Transkribus crowd for their great work! An overview of all public models you can find here or directly in Transkribus (via “Tools”-tab -> “Models”).
