Some guys here use Spanish and Mandarin phonemes for building German words. How can this be achieved?
2 Likes
For my part i am using Spanish phonemes for a song in Latin, but since i can do it all in Spanish phonemes, i just set my whole track to Spanish and type in the phonemes manually, for instance for “draco volans per caelum” i would enter “.d r a k o | .B o l a n s | .p e r | .k a e l u m”.
I didn’t have to use cross language yet, but i will in a future song written in a made-up fantasy language, so i suppose the process will be similar than for other real world languages that require phonemes from multiple languages. I suppose in these case we have to set different languages on different notes in order to use different phoneme sets?
Since you can write phonemes from different languages per note, Use the XLS feature to manually enter notes in the language closest to the original language. among the languages supported so far. Phoneme length/intensity adjustment and Mouth Opening are also useful for singing in languages that are not fully implemented.
I implemented Korean this way when it wasn’t supported in SV1 and it worked pretty well.
+I use translator when I write more than 2 lines in English. I’m not fluent that, so if you don’t understand something because the sentence is strange, please ask me again.
What is XLS feature ?
XLS = Cross Language Sythesis.
In the Voice panel you can select a base language that applies to the current group, while in the Note panel you have a “language” drop-down to override the language for that note, where you can select another language.
1 Like
i would suggest .U o l a n s not .B o l a n s
unless you’re aiming for ecclesiastical pronunciation, but then, it would be .ch e l u m not .k a e l u m
so either use .U o l a n s and .k a e l u m, or .B o l a n s and .ch e l u m. Be consistent! And I would suggest using the Mandarin :n to get a softer n at the end of a syllable before another cosonant, as in .U o l a :n s (.U o in Spanish, l a :n s in Mandarin)
Thanks for the suggestion, I’ll try that. You’re right, there are multiple different Latin pronuntiations, and the Suno-generated song I am using as reference is not consistant, it is mixing up classical and ecclesiastical pronuntiation. In classical Latin indeed the letter written as V was actually a U.
I did notice it mixed both pronuntiations, but as a starting point I made it the same as the Suno song, and I am planning a pass to fix these inconsistencies, havn’t gottent to that step yet. I am still in the process of going over the translation mistakes. I don’t know Latin myself (although I can understand some words since I’m french speaking and many french words come from Latin), had help from an older ChatGPT model to translate, but the newer one is better at it!
After several iterations of validating spelling, syntax, grammar and accuracy of the meaning, it’s getting closer and closer to a correct version. Then I’ll work on fixing the pronuntiation.
I regularly cover songs using made-up fantasy languages, and I do use the approach of using different languages on each note, according to what that word needs.