Making a diffsinger. It's fun.

[Suzuki Hoshi]

I'm labelling a CVVC (only using of 1 of 7 pitches) that i originally made for DeepVocal
I hope the colab part goes well but idk if it will

 

[heynotloid]

wait, you're porting the CVVC (ka-ke-ki-ko-ku-ka, you know) files to deepvocal?

 

[Suzuki Hoshi]

wait, you're porting the CVVC (ka-ke-ki-ko-ku-ka, you know) files to deepvocal?

i was but now i'm doing a diffsinger
this is my only problem so far

 

[heynotloid]

i was but now i'm doing a diffsinger
this is my only problem so far
View attachment 13502

diffsinger does not work using the samples of a normal UTAU like cvvc vcv and cv, to create a diffsinger (well record it) you must sing a song acapella, if you want to add appends, you should sing them with that tone, usually people record 1-3 hours, 2 hours are the ones that ensure a better result, also several languages are recorded (in some cases) using recordings in UTAU formats is not bad, but it is not recommended, since singing is different from recording for UTAU, and what diffsinger needs is singing

Post automatically merged: Friday at 6:18 PM

diffsinger does not work using the samples of a normal UTAU like cvvc vcv and cv, to create a diffsinger (well record it) you must sing a song acapella, if you want to add appends, you should sing them with that tone, usually people record 1-3 hours, 2 hours are the ones that ensure a better result, also several languages are recorded (in some cases) using recordings in UTAU formats is not bad, but it is not recommended, since singing is different from recording for UTAU, and what diffsinger needs is singing

my examples

Post automatically merged: Friday at 6:21 PM

(i need edit the samples because the background sounds aja)