So for the past while I’ve been working on a CV voicebank for PhaseU.
Currently it is recorded in three octaves, and as such the vocal range is pretty large. I’m not ready to release it quite yet, but it should be ready soon enough.

The sample below is a demonstration of the capabilities. As with most voicebanks you can genderbend it in order to make it sound female/male. Thanks to the C4 samples UTAU can render vocals up to around C5 before it sounds too squeaky (as long as you use a g-10 to g-20 flag).

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

Helloes!

I’m taking another short break from UTAU and stuff. My PC seems to be leaking electricity into my soundcard, making it hard to record samples. Although I’ve managed to remove most static, I am looking for a better solution. It is possible to remove nearly all the static, but I like having high-quality samples.

If you listen closely to the CV voice bank you’ll hear static from the samples. So yes, I have to re-record the entire CV bank.

On the other hand, I couldn’t stop being curious, so I tried recording samples of my deeper voice (as in, not the one I talk in, but the one I goof around in), and it actually came out rather OK, surprisingly enough.
It may be a future project.

Here’s a test sample. They’re really low (A2) but sounds a lot deeper than the regular PhaseU bank. I also placed them a bit low on the piano roll, the first try sounded a bit like a cello, a talking cello.

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

Speaking of PhaseU, I’ve decided that I’m re-recording some of the samples due to consistency problems. Maybe I could utilize my MacBook for this though.

I’ve finally finished up PhaseU to a state I’m comfortable with releasing.
Do note that this is not a final release, it’s merely a public beta release. There are a few samples that aren’t the way I want them to be, and I’m sure there’s a few samples in there that might be misnamed or incorrectly configured.

However, due to the sheer amount of syllables available, it would take me forever to test it out all on my own, so I set up a bug reporting page found here so others can report bugs they find in the voicebank.
If there’s a sample missing (not too likely with the VCV part of the bank, but I didn’t use a reclist for the CV part, so there may be a sample or two missing) or in-case someone would want to request a sample, they can do that on the same bug reporting page.

The download, samples and technical specifications can be found on the PhaseU project page.

Hello!

The past few days I haven’t been working much with the voice bank. Instead I’ve been focusing more on my personal life as well as on Megaten. I have switched out a few samples that irk me, fixed a few, and I’m probably going to switch out a few more samples before I’m completely happy with the VCV part.
However I grew tired of the VCV part of the bank a while back, hence why I’ve been taking a slight break from it.

Today I recorded an additional 30 samples, similar to VCV samples, but these will be utilized for CV and VC capabilities (including L syllables).

The CV bank has a total of 18 samples (five syllables per sample). And the VC bank has 5 samples (I think at most one sample has 16 syllables).

Since I’m doing my best to keep the amount of samples down a bit, the CV and VC part of the voice bank sadly will not support Latin characters either (save for the L samples).
The voice bank will still (mainly) be aliased in kanas.

I decided to add VC samples as well in order to make things like “masu” or “desu” be pronounced easier. As most people know, in some cases “desu” and “masu” has a quiet u on the end, so combining the syllables “de” and “es” will give you a “des” sound.

As usual with CV banks, it sounds a lot more robotic than the VCV bank, but I figure the more capability the merrier.

Lastly, I started recording a few extra samples as well. This include English pronounced “the thi thu tha” sounds, three breathing (in) phonemes, and English counting, 0 through 10.
I may come up with more extra samples eventually.