VCV English

[PeyPeytonton]

Yes, the fabled old wife's tale is possible (Adrian is also VCV English I believe //shot) because I am a masochist when it comes to recording. (The reclist is a mess, but I'm trying to clean it up.)

It basically gives the same results as VCCV, if not a little more flow-y I guess? But let me know what y'all think.

Sample:

Spoiler

Sample Vocals Only:

Spoiler

Critique please. :D
I want this to sound as good as possible. //shot

 

[na4a4a]

VCV english generally isn't practical tbh. You need a mix of VCV and CVVC to still get all the sound combinations you need, along with the bank being very large. I actually am working on an english bank with VCV added to it on a practical level.

How did you record the samples? did you record samples like "ast ka"? in which case it's impossible because a full bank wouldn't fit in utau lol.

 

[PeyPeytonton]

I accept that a full bank isn't really possible. This voicebank itself is just 85-ish% - CV/VCV/VV while the rest are just VC's. Doing everything would be true VCV, but yeah, that's kind of illogical and I would run out of OTO lines realllll quick.

(I recorded the samples like japanese VCV [bla_blI_bl{_and so on] and then [n_ka] [m_ka] [l_ka] [ng_ka] etc.)

But yeah, thank you and stuff for responding~

 

[Kiyoteru]

I've written VCV english reclists a couple times, and there's a method of organization that I generally find to be the most sensible.
For the VCV strings itself, just the solo vowels and solo consonants. No diphthongs, because you can fake them using the transitions. For example, "my mind" would be [- ma] [I ma] [I nd].
With syllable-initial clusters, many can be constructed from putting a CC and CV together. I recommend having an L vowel (like, the L version of the 3 vowel) so that you can do things like [bul] [la] for "bla". The ones that can't be done this way are st, sk, sp, sn, and sm, which is few enough that they can just be in the regular VCV strings.
For syllable-final clusters, plain CVVC-style recordings are good.
You have about 32000 or so lines of OTO space. I did the calculations for my most recent list and I could just about fit 4 pitches in.

I agree that Adrian is a really fun voicebank to work with! But it's more trilingual than english specific... and it would be fun to have other voice types to use.

 

[PeyPeytonton]

I've written VCV english reclists a couple times, and there's a method of organization that I generally find to be the most sensible.
For the VCV strings itself, just the solo vowels and solo consonants. No diphthongs, because you can fake them using the transitions. For example, "my mind" would be [- ma] [I ma] [I nd].
With syllable-initial clusters, many can be constructed from putting a CC and CV together. I recommend having an L vowel (like, the L version of the 3 vowel) so that you can do things like [bul] [la] for "bla". The ones that can't be done this way are st, sk, sp, sn, and sm, which is few enough that they can just be in the regular VCV strings.
For syllable-final clusters, plain CVVC-style recordings are good.
You have about 32000 or so lines of OTO space. I did the calculations for my most recent list and I could just about fit 4 pitches in.

I agree that Adrian is a really fun voicebank to work with! But it's more trilingual than english specific... and it would be fun to have other voice types to use.

Yeah I don't really have any dipthongs in there. I would just do [- ma][a i][i ma][a i][i n][n d] for My Mind. I split everything up while still making it sound smooth so I didn't have to do soooo many recordings.

I recommend actually recording all the [bla] [bra] [kwa] sounds because it makes it smoother but that's just me.
But that's kind of interesting. I may do sort of a "LITE" list.


Adrian is made to sing every language btw. I just said English because they're like the only VCV english example ever.

Thank you for replying. :^D

 

[Chianachini]

Your samples are very smooth by the way~ I wouldn't be able to tell a difference!

I've experimented with 100% English VCV recordings when waiting for VCCV to come out, recording ending consonants as VCV, but still oto-ing like you would VC (as I imaged would be the concept of VCCV). The results were ok, but still very awkward. Some samples just don't sound natural when recorded in long strings and deserve special recording styles.

We wouldn't be making progress if we didn't experiment! It wouldn't be very fun if everyone used the same recording standard, so it's up to the individual to try things out and see what's practical for them. They may even discover something that's useful for the whole community while they're at it!

 

[PeyPeytonton]

Your samples are very smooth by the way~ I wouldn't be able to tell a difference!

I've experimented with 100% English VCV recordings when waiting for VCCV to come out, recording ending consonants as VCV, but still oto-ing like you would VC (as I imaged would be the concept of VCCV). The results were ok, but still very awkward. Some samples just don't sound natural when recorded in long strings and deserve special recording styles.

We wouldn't be making progress if we didn't experiment! It wouldn't be very fun if everyone used the same recording standard, so it's up to the individual to try things out and see what's practical for them. They may even discover something that's useful for the whole community while they're at it!

Thank you. :DD

Yeah it would get pretty boring if everybody uses the same reclist. //shot

That's true. If experimentation didn't exist, English, and many other languages on UTAU wouldn't exist.

 

[PeyPeytonton]

Progress is halted because I'm in the process of cleaning up the reclist, adding a few more sounds, and then I need to re-oto it.