Hi,

for the record, using one core japanese voices, it seems it doesn't read the lines if automatic language switching is on, but it does read them properly if it is off.

-- Patrick

Le 27/07/2022 à 21:37, Sharni-Lee Ward a écrit :

I saw a post on one of my friends' Tumblr blogs with two short passages of Japanese in it. I took the opportunity presented to me to test out the Japanese 1Core voices I had installed.


I tested with all three of them (they probably use the same logic, so I didn't necessarily have to do that I suppose), and was soundly disappointed by the results. Espeak identified the strings of characters as Japanese-Letters. When I read the same text with the 1Core voices meant to /read/ Japanese, I got /silence./ It read the Latin characters in the distorted way I had expected, although NVDA switched to Microsoft David when reading actual English words and sentences. The Japanese passages, however, were simply treated like lines of unidentifyable/unassigned characters, with complete silence. I could not even scroll over them character by character with the left and right arrows and get a response from them!


For reference, this is the post I discovered this problem with:


https://nam12.safelinks.protection.outlook.com/?url=https%3A%2F%2Fkintatsujo.tumblr.com%2Fpost%2F690881533161144321&;data=05%7C01%7C%7C8327346985aa461f911f08da705aa7aa%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C637945831927015770%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=mCbYNFHdM%2F7SYkNGN541KfULVmmXUsYFP%2FpUrVjRx8A%3D&reserved=0


I do not need or want a translation, I just want this behaviour verified by someone else.