MBROLA voices - no sound about realtimesingingsynthesizer HOT 3 OPEN

Hi!

I also failed trying MBROLA in the past unfortunately and I'm too lazy to debug C code right now.

About the notes per syllable it should be doing that. Long words should be splitted into syllables if you have many notes to sing. You can try visualizing the musicxml file with musescore to check it. Let me know if you find some issue with that.

from realtimesingingsynthesizer.

Thanks for your answer, @mathigatti, I can understand the situation with the MBROLA voices.

About the long words...
Can you suggest a way, how I can split them into syllables, and still keep the original sounding of the words?

Thanks!

from realtimesingingsynthesizer.

The software should do that for your automatically, let me know if there is something specific you want to do that's not working and I will give you an example of how to do it

I mean if you write "artificial" and you want it to sing 4 notes, it's going to split it into "ar" "ti" "fi" "cial". If you want it to sing 6 notes it will do something like "ar" "ti" "fi" "cia" "aa" "al"

Anyway I'm testing it and it looks like, as you say, it doesn't keep the original sounding of the words. You can check this other project which has a slower synthesis time but better singing quality

from realtimesingingsynthesizer.