I've trained a G2P model of written eastern Norwegian using the train.py file included in OpenUtau. Clap for my thin HP laptop with integrated graphics that SOMEHOW did not overheat!
Now, I'm waiting for Visual Studio 2022 to install so I can continue the work. Training the model was easy, actually making the phonemizer is up next... That'll be interesting!
I haven't included a dict.txt yet, since I don't know what (if anything) the phonemizer might pronounce wrong. I'm not good enough at exceptions with words yet.
I can't remember if I said it kast log, but I finshed the first draft of the reclist. It doesn't include sounds not found in eastern ("standard") Norwegian though like, uh, tjukk L, palatalisering, skarre R... If needed, it will probably be possible to make adjusted reclists and phonemizers to accomedate this.
Comments
Displaying 0 of 0 comments ( View all | Add Comment )