A short one - but I’ve noticed from my first batch of transcriptions (a 270 page diary written in English and dated 1822) that Leo struggles to identify semicolons. They are - in my opinion - quite obvious in the text, but they are transcribed as commas every time.
So sorry for the delay in getting back to you on this. I have attached two photos showing semicolons the software hasn’t recognised - they are from two different collections. I’ve also included pictures of a couple of other small things, as follows:
sometimes it doesn’t seem to be able to recognise eighteenth-century variations of the ampersand;
for some reason it really struggled to pick up on the signature of one of my writer’s - Jeff[rey] Amherst - it transcribes it differently each time, but never correctly;
the software struggles to recognise Indian place names (or more accurately, Anglicised versions of Indian place names). This example is for the place ‘Bhurtpore’.
I hope all of this is helpful and sorry again for the delay.