I need help in knowing details to how to make a voice model

[mona1222002]

Hi, I got introduced to nnsvs after trying talknet and asking if there is any way to perform voice synthesis and get a better quality, but I actually need some help in knowing how to make the dataset, so if you can help me by giving me some detailed instructions and telling me what softwares to use to help in creting the dataset, I'll be really thankful

[mona1222002]

Hi, I got introduced to nnsvs after trying talknet and asking if there is any way to perform voice synthesis and get a better quality, but I actually need some help in knowing how to make the dataset, so if you can help me by giving me some detailed instructions and telling me what softwares to use to help in creting the dataset, I'll be really thankful

and can I have English and Korean together in the same dataset and train them together

[oatsu-gh]

Hello,

asking if there is any way to perform voice synthesis and get a better quality

More amount of sample of voices (or songs) that have consistent singing style will provide better results. Less-noise WAVs are appreciated.

I actually need some help in knowing how to make the dataset

The following document will be helpful to make datasets. https://docs.google.com/document/d/1uMsepxbdUW65PfIWL1pt2OM6ZKa5ybTTJOpZ733Ht6s/edit

what softwares to use to help in creting the dataset

Score: UTAU for UST file / MuseScore for musicxml file. LAB: wavesurfer, audacity or oto2lab WAV: any DAW you like

can I have English and Korean together in the same dataset and train them together

Yes but so complicated that I recommend making normal small one at first.