This really came as a problem when I tried to use the four languages available for my soundbank and cross to Neapolitan (usually regarded as a “dialect” of Italian but is actually another language with a close relationship). And since the Neapolitan phonology is different from all four languages, I have to do a lot of cross-languaging (mostly between English and Cantonese, and a bit of Japanese). However, I gave up when I found it such a horrible task to do… The same phone in different languages are coded differently and different phones in some languages are coded the same, which is hard to see when phonetic symbols from all different languages. Well, basically it’s because the two Chinese languages use IPA, but Japanese uses romaji and English uses arpabet… I wonder if this can be just all IPA in future versions… Also I wonder if phone-level cross languaging is possible
I don’t know if this is what you mean by phone-level cross language - I use one note for one phone with a dot prefix in lyrics and then I can set different language for every note/phone. I also can control lenght of phones independenly.
I tried to make a transcription script from IPA to synthV’s X-SAMPA / ROMAJI / ARPABET with automatic language assignment for notes. It worked for German somehow, but some phonemes are missing anyway.
The transcription is a part of my RVnotesFromTextgrid script, which is not yet published.
Will be in a new update.
This is an example of Nena - Leuchtturm with SVP link.
It is Tsuina chan multilingual (English + Japanese only) copy of original with RV scripts.
(I’m sorry, the first one was not Tsuina, now it is right)
This is how it looks in Praat after alignment with Montreal Forced Aligner and manual corrections: