Skip to main content

RECLIST. UPDATE. THREE. BABEY!

Anyways, I guess I technically tested it out and it works. Thanks guys. But I have like, a few notes on this, obviously.

I added a bunch of stuff it didn't need.

Okay so some bad and good news. I'll start with the bad news:

I added a bunch of stuff that I didn't really need to and it expanded the size a whole bunch. This wouldn't be bad if not for the fact that I made this purely with the intent of getting around the time + oto limit constraints. Unfortunately, this means that I am going to have to spend a bunch of time going back and removing all that stuff. But with that comes good news as well.

Now knowing I have a lot less work on my hands, I think it'll be a lot faster to make. On top of that, I realized that I could take my other method, the semi-VCV with VC additions, and just remove actual VC configs and it sounds, like, the same, in a freaky kind of way.

Most of all, I made a wild realization:

I didn't need a full set of VCs

You only need them when you don't have a full set of vowel transitions with this method. Since my method pretty much relies on having a full set of transitions to sound good, I'm genuinely not sure why I even tried to add these in.

So that answers one of my big questions from the original post. I was really just so used to CVVC that I forgot that I wasn't trying to write CVVC.

To give a pretty good example as to what's up with this whole feature, previously, you might write out "starlight" like this:

[sta] [a r] [la] [_ ai] [_t-]

This doesn't make a ton of sense, because you technically already have the [_ 3], [_r] and the [_ a3] transitions. It would sound much better if you used the [_ a3] transition, giving you this:

[sta] [_ a3] [la] [_ ai] [_t-]

If you wanted to naturally extend the "l" sound, you also have some options outside of [3 l] that'll work a lot better with this method. You have [_l] (20ms overlap VC transition) and [_ l] (25 preutterance 50 overlap solid vowel transition), giving you something like this:

[sta] [_ a3] [_l] [la] [_ ai] [_t-]

Potential notation changes

Also, to note, I'm working on changing the notations for some things. I've currently got the vowel to diphthong vowel transition as an underscore followed by a space, while the vowel to consonant is just an underscore (changing [_ai] to [_ ai], etc.). I'm gonna change this again, because it was a last minute resort for VC 20 ms ovl l / r / m sounds being overridden by VV 25 pre / 50 ovl transitions. The current likely candidate is using an asterisk to denote vowel continuation, giving you [*l] instead of [_ l], [*ai] instead of [_ ai], and [*i] instead of [_ i]. I particularly like this one because it was an attempted standard back in like 2015, it was easy enough for me to remember for 5 years despite having depression-induced amnesia.

On a completely unrelated note...

If you are solely interested in reclist stuff I'd recommend stopping here. tl;dr I made an UTAU plugin and I found a major bug (if you're using a Japanese-encoded vb).

So I have also been making this UTAU plugin that basically parameterizes a bunch of stuff and lets you save presets. So far it just does, like, pitch bend randomization and vibrato. It's not terribly far along.

I found a bug that has to do with encoding, it can't rewrite the lyrics property if it's in Japanese, which would be a huge issue if I cared much about this plugin or if people used it, but I really don't care and I'm the only person who uses it so nbd, I'll just fix it when I'm done with the reclists or if I get bored.

To be a little more specific, I don't care much about this atm because I plan on writing my own ini parser / writer specifically for UTAU-flavor ini files, because I have to do an absurd amount of work just to get the one I'm using to play nice with UTAU (AND I need this library specifically for another project I'm working on that I'd rather not talk about yet). It's not only a lot of work on its own, but it's a massive change when I do end up replacing the library in my plugin. I've learned that sometimes it's nice to not sweat it too much with personal projects :)

Comments

Popular posts from this blog

Generic Yearly Life Update

It's not, like, terribly unusual for me to disappear for months \ years at a time on here, but I thought I'd give at least a little update as to what I've been doing and what I have NOT been doing because I don't really want to leave anyone hanging, in the event that anyone has been keeping up with my posts.

Shhh... it's a secret.... here's a Carmine Mini 1 prerelease distribution.......

* Hands you a flash drive * you didn't get this from me. Same rules apply as with Urin Sasaki 3. Use like Urin Sasaki 3. The vb should be in the "Carmine" folder. Fully otoed but like 90% untested. Works best with TIPS  and most WORLD-flavored resamplers. Many known issues in recordings -- recorded in under 10 minutes probably. No audio samples we die like men. Also I'm in the middle of doing some finalization of that dumbshit 5 pitch voicebank or whatever, and more reclist testing (english stuff ya know), I promise I didn't die, just really busy. Byeee

Reclist update four???

I am mostly doing finalization stuff. It's weird to say because I've been working on these two methods for over 5 years (I need to do normal things with my life for once...). This'll mostly go over some minor (major in terms of workload) standards changes and the planned addition of missing items, which there are many of. After all this, it'll just be finalizing the configs, documentation writing and release. And I added a sample :) Standard changes First off, I made a change to the vowel transition standard, which was originally [_Au], changed recently to [_ Au], and is now [*Au]. But because of a more recent issue I am having, I also wanted to change the vowel ending standard. However, it's a hard issue because the most direct, logical course of action would be to change [A -] to [A-]. I'm wanting to do this for compatibility with a plugin, and it's unfortunately a pretty annoying and very nonstandard change compared to my other reclists (which matters...