Could you please change the Japanese script?

[shigekose]

Hello! I am Japanese and a user of your wonderful model. However, the Japanese script for this piper-recording-studio was unable to output the 'dataset.jsonl' with the correct phonemes using 'ljspeech' during 'piper''s training. This issue stems from the fact that Japanese contains three types of characters. One of these, 'Kanji', is recognized as Chinese in 'ljspeech'. The pronunciation of Japanese 'Kanji' is entirely different from Chinese characters. Therefore, when outputting the 'dataset.jsonl', it is necessary to use 'Katakana' or 'Hiragana', which reflect the original Japanese pronunciation. I have trained 'piper' using metadata I created myself. Please feel free to use this. Additionally, I will attach the URL of the Japanese corpus I used.
I apologize for the awkward English as I used a translation tool.

metadata.csv

https://github.com/mmorise/ita-corpus

Thank you for your excellent model.

Best regards,
Koh-say Shigueta