For all those who need to hear this: Together with two colleagues, Javier de la Rosa (@versae) and Álvaro Cuellar, I created a small language model (sequence-to-sequence) to modernize non-normalized 17th century Spanish texts.
A short introduction to it can be found here:
https://dh2022.dhii.asia/abstracts/files/DE_LA_ROSA_Javier_The_Moderni_a_Project__Orthographic_Modern.html
and the model itself can be found here on Huggingface:
https://huggingface.co/spaces/versae/modernisa
The basic idea for this project is that you definitely need normalized texts to conduct computational analyses.