Jinriki UTAU help? What can I do to make my UTAU sound more like the voice samples?

WannabeHistorian604

Teto's Territory
So I'm trying to make an UTAU of this character. I've gotten pretty far, and I've tried a few covers with him. The character is Jekyll from the Fate series, so thinking I was being clever I decided to cover Jekyll&Hyde with him.

This is the character's original voice:

The UTAU kind of sounds like him but is it just me or does he sound... What do I say, hoarse? Muffled? What can I do to make him sound more like the target voice? Is it tuning? Is it oto-ing? Is it flags? Is it resamplers? Or should I change the samples entirely?
Also, the funny thing is, he appears to sound somewhat different when I use this ust: What's going on here?
I think that this Nico Nico user has gotten closer to my target voice: Is this just a voice sample thing or is there some kind of technique to make the UTAU sound like this?
Hopefully this doesn't count as distributing the voicebank!
 
Last edited:

Halo

Icon by Wanpuccino @ DA
Administrator
Defender of Defoko
+ Clean, decent samples (as close to a long, held note as possible, similar pitch if possible, obviously in this case that is pretty difficult, but still. aim for the moon etc)
+ Good oto
+ Make sure to set the UST; remove existing preutt, overlap and STP from all notes. This + flags probably explain the difference between the two USTs
+ Pick a good resampler for the voice. Try all of them you have access to, its largely personal preference. The example you posted sounds like TIPs or tnfnds to me, but maybe doppeltler or moresampler.
+ Flags might help get the voice you want, but won't fix wonky sampling or otoing.
+ Good tuning will always help.

Some notes in approx order of importance
 

WannabeHistorian604

Teto's Territory
Thread starter
+ Clean, decent samples (as close to a long, held note as possible, similar pitch if possible, obviously in this case that is pretty difficult, but still. aim for the moon etc)
+ Good oto
+ Make sure to set the UST; remove existing preutt, overlap and STP from all notes. This + flags probably explain the difference between the two USTs
+ Pick a good resampler for the voice. Try all of them you have access to, its largely personal preference. The example you posted sounds like TIPs or tnfnds to me, but maybe doppeltler or moresampler.
+ Flags might help get the voice you want, but won't fix wonky sampling or otoing.
+ Good tuning will always help.

Some notes in approx order of importance
When I get permission from the original voice actor (I at least emailed the fanclub, the closest thing to the actor I can get), can I get help with the otos?
I tried different resamplers, but for some reason I can't get rid of the voice losing some of its timbre for some reason.
 

Halo

Icon by Wanpuccino @ DA
Administrator
Defender of Defoko
Gonna be real, you're likely not going to get permission unless you're a company looking to hire him, and even then for an UTAU? That's... not likely. You're better off trading in the murky waters of fan appreciation & quiet trading on niconico/twitter and discord groups. It's not really harmful unless you publicise it or it gets really popular, so honestly its probably better to not bring it up with anyone official in the future to avoid bringing undue attention to the Japanese members of the community, at least when it's a known seiyuu with a fanclub and all.

UTAU will always degrade or change voice timbre, you can only play roulette with resamplers and the best samples you have unfortunately :sad: You can recreate some of it with flags and like i said better samples minimise it.

As for help, I will always respond to check otos visually or explain what's happening in a broken audio sample, but for a tutorial its good to check these:
 

WannabeHistorian604

Teto's Territory
Thread starter
Gonna be real, you're likely not going to get permission unless you're a company looking to hire him, and even then for an UTAU? That's... not likely. You're better off trading in the murky waters of fan appreciation & quiet trading on niconico/twitter and discord groups. It's not really harmful unless you publicise it or it gets really popular, so honestly its probably better to not bring it up with anyone official in the future to avoid bringing undue attention to the Japanese members of the community, at least when it's a known seiyuu with a fanclub and all.

UTAU will always degrade or change voice timbre, you can only play roulette with resamplers and the best samples you have unfortunately :sad: You can recreate some of it with flags and like i said better samples minimise it.

As for help, I will always respond to check otos visually or explain what's happening in a broken audio sample, but for a tutorial its good to check these:
Will you check the otos for me? I don't know if that will fix my issues, but it's worth a try.
 
Last edited:

Halo

Icon by Wanpuccino @ DA
Administrator
Defender of Defoko
You know, usually I wouldn't and I would encourage you to post pictures of the waveforms in this thread, but if you promise to listen to me explaining if anything is wrong and not rely on me to take the time to fix it myself, then I'll happily take a look at the bank privately. DM me! @WannabeHistorian604
 

Similar threads