• If you do not recieve your confirmation email within a few hours, please email haloutau@gmail.com with your username for manual validation. Your account should be activated within 24 hours.
    You may also reach out via any other listed contact on Admin Halo's about page: https://utaforum.net/members/halo.194/#about

Open Other Can somebody do the Google Colab part of the DiffSinger training for me?

heynotloid

Momo's Minion
I can't do it myself (it doesn't work on my computer for some reason)
On the diffsinger discord you can find people who do training from 10 dollars, since it is quite a laborious job. If I knew how to do it I would do it, but unfortunately I don't. And how many hours of recording do you have?
 

Suzuki Hoshi

Ruko's Ruffians
Defender of Defoko
Thread starter
On the diffsinger discord you can find people who do training from 10 dollars, since it is quite a laborious job. If I knew how to do it I would do it, but unfortunately I don't. And how many hours of recording do you have?
my mom won't let me buy things like that
I have two versions; one that's already labelled (basically a Kyoune Logan (SUZUKI HOSHI'S NEW NAME) CVVC JA) and one that isn't (about 10-15 minutes of data)
I'm going to make a new (English) one and I may need help labelling.
 

heynotloid

Momo's Minion
my mom won't let me buy things like that
I have two versions; one that's already labelled (basically a Kyoune Logan (SUZUKI HOSHI'S NEW NAME) CVVC JA) and one that isn't (about 10-15 minutes of data)
I'm going to make a new (English) one and I may need help labelling.
15 minutes is very little time, you need at least an hour of recording to generate a good result, with 15 minutes the voicebank will sound very bad and will have many bugs, also I think I mentioned that you don't have to use the UTAU phonemes, but recordings of you singing.
 

Suzuki Hoshi

Ruko's Ruffians
Defender of Defoko
Thread starter
15 minutes is very little time, you need at least an hour of recording to generate a good result, with 15 minutes the voicebank will sound very bad and will have many bugs, also I think I mentioned that you don't have to use the UTAU phonemes, but recordings of you singing.
i'm going to sing now (English songs only.)
 

heynotloid

Momo's Minion
15 minutes is very little time, you need at least an hour of recording to generate a good result, with 15 minutes the voicebank will sound very bad and will have many bugs, also I think I mentioned that you don't have to use the UTAU phonemes, but recordings of you singing.
If you recorded 15 minutes for an append, I think it's fine, but if it's for the entire voicebank, it's going to give you a lot of problems.
 

Suzuki Hoshi

Ruko's Ruffians
Defender of Defoko
Thread starter
If you recorded 15 minutes for an append, I think it's fine, but if it's for the entire voicebank, it's going to give you a lot of problems.
okay! I'm going to record now, as I said, but feel free to say something (it might be an hour or two before I check it tho)
 

Suzuki Hoshi

Ruko's Ruffians
Defender of Defoko
Thread starter
Did you sing the audios in Japanese? And I recommend that if you're going to make a Diffsinger voice bank ONLY in Japanese, you record for an hour or a little more.
I'm going to make an English Diffsinger RN but I did make a JP one at one point that's never going to be released