UTAU + VOICEROID Software?

[Mei-Saime]

Recently and currently, I'm asking for people to help me test a speaking voicebank for Yuzuko here
http://utaforum.net/threads/yuzuko-exvoice-beta-testers-needed.17243/
But maybe if this goes well, could we have an idea of a UTAU text-to-speech program like VOICEROID?
Plugins are nice, but sometimes they could be difficult to use so MillyAqualine and myself were talking about some ideas for a UTAU text-to-speech program and if anyone was interested to help make it with Yuzuko's beta speech voicebank. Leave your thoughts and if interested comments here or PM me.

 

[Sors]

This would certainly be interesting! As someone who wants to work on an UTAU based Visual Novel, text to speech would be a very nice addition instead of using talkloids in case voice providers were unable to voice act.

 

[Mei-Saime]

This would certainly be interesting! As someone who wants to work on an UTAU based Visual Novel, text to speech would be a very nice addition instead of using talkloids in case voice providers were unable to voice act.

True! After making this speaking voicebank, maybe this could make some people consider a UTAU text-to-speech software! ^^

 

[Info-Chan]

As far as Japanese UTAU TTS, we do have UtaYomi which loads UTAU vbs and turns them into TTS voices. For English, it's more complicated, as you can't really do such it requires a dictionary.

 

[Mei-Saime]

As far as Japanese UTAU TTS, we do have UtaYomi which loads UTAU vbs and turns them into TTS voices. For English, it's more complicated, as you can't really do such it requires a dictionary.

That's true but then again if you only have one pitch on that certain vb, then it sounds like they are singing, which leads to more fixing of their TTS vb to be improved for speech.

 

[Info-Chan]

Actually, that's not true. UtaYomi uses the UTAU engine to add pitchbends into speech, so monotone samples don't really affect it.

In other, actual TTS programs, monotone samples do affect the output resulting in singing like speech, but you do not need extra pitches and in fact most TTS programs don't use multipitch samples as for most of them its not possible or pointless.

 

[Mei-Saime]

Actually, that's not true. UtaYomi uses the UTAU engine to add pitchbends into speech, so monotone samples don't really affect it.

In other, actual TTS programs, monotone samples do affect the output resulting in singing like speech, but you do not need extra pitches and in fact most TTS programs don't use multipitch samples as for most of them its not possible or pointless.

That's true.

 

[Info-Chan]

That's true.

If you truly would like to have TTS for your UTAU, you'll need to research speech synthesis as it's very different from singing synthesis. You may have all the transitions in an UTAU vb, but you need to label them, compile the voice, and run it in a TTS. Most use Linux as the OS too. I'm not trying to be a debbie downer or anything, I'm simply trying to be realistic.

 

[Mei-Saime]

If you truly would like to have TTS for your UTAU, you'll need to research speech synthesis as it's very different from singing synthesis. You may have all the transitions in an UTAU vb, but you need to label them, compile the voice, and run it in a TTS. Most use Linux as the OS too. I'm not trying to be a debbie downer or anything, I'm simply trying to be realistic.

I know. I was just looking for ideas and stuff. ^^