UTSU - A cross-platform vocal synth frontend

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
If at all possible, when you start the oto part of develop I highly recommend the option for an extra set of overlap and consonant lines.. (the red and green ones lol) so you can have consonants at the end of samples n stuff... NIAONiao is this way.

Please I beg of you
This is just a frontend made to work with existing UTAU resamplers, so I don't think that'd be possible without making a new resampler too ^^;;
 

Shasta

Professional Lurker
Global Mod
Defender of Defoko
I did a bit more testing with it, and I noticed a few things with the utau rest notes and utsu's lack of such. This could be part of the issue with the utsu's inability to open usts. I can open a ust made in utsu in utau, but not the other way around, and when I do open an utsu ust in utau, the rests are just longer notes.
untitled_drawing_by_shastaisanandroid-dc08owu.png
untitled_drawing_by_shastaisanandroid-dc08ojw.png


Edit: I use a windows 10 laptop.

Double edit: I'm currently looking at the difference between an utsu ust and an utau ust to see if they're written differently in hopes it'll help. I'll tell you what I find.
 

수연 <Suyeon>

Your friendly neighborhood koreaboo trash
Supporter
Defender of Defoko
I wish I could try this for myself (I just want to see if non-Japanese scripts like hangul work), but unfortunately, it won't even open after installation.

error.png error2.png

I'm using the latest version of Windows 10 and I received this error for both the version that needs javascript before install and the one without javascript (when I had java, it was confirmed to be the latest version upon checking for updates).
 
  • Like
Reactions: partial

Shasta

Professional Lurker
Global Mod
Defender of Defoko
Idk if any of this information will be beneficial, but from what I am seeing so far...

If you open an utau ust in notepad, and an utsu ust in notepad, they look similar, but read different things. On the left is the utau version of the ust and on the right is the utsu version.

Few differences I noticed, first in the [#Version] section is obviously the difference in the versions. Then in [#Setting] I noticed the ust opened in utau does not have the time signature, but has the track number and a place for cache'd files. It also lists the resamplers I prefer to use (Which is a personalized setting I have for my editor..)
bleg_by_shastaisanandroid-dc08ql4.png


The biggest difference I noticed is with the notes themselves, which is the [#0000] etc ones.
I took a closer look at the area between #0002 and #0003, which would be the space/rest between the third and fourth note, and this is what I saw.
bleg2_by_shastaisanandroid-dc08rzy.png

Basically, they look the same. So I went back and edited the ust in utau to give it the rest/space it's supposed to have like it does in utsu and this is what I saw.
upload_2018-1-16_21-36-13.png
The one on the left is the utau ust without the rest/space, and the one on the right is the one with the rest/space. It reduced note #0002's length down to 480, rather than 960 like it is in the first utau one and the utsu one. From what I am seeing, duration=/=length. Length is how long the note is, and duration is how long the note is actual audible. In utau, we don't have duration in our usts. So I see this posing a problem with the cross compatibility of utsu usts and utau usts.

Again, I have no idea how useful any of this information is, I'm not a developer and I know nothing about stuff like this. I'm just trying to help to the best of my abilities.
 
  • Like
Reactions: Nohkara and Lethe

Lethe

Ruko's Ruffians
Supporter
Defender of Defoko
Thread starter
aaaaaaa so many Windows bugs ;w;

I have a Windows partition on my old laptop--I'll boot that up right now and see if I can debug all these issues. In the meantime:

Few differences I noticed, first in the [#Version] section is obviously the difference in the versions.
2.0 is the UTAU-Synth version, while 1.2 is the Windows UTAU version. When saving a file, you should get 3 different file formats to choose from:
ust 2.0 (Unicode)
ust 2.0 (Shift-JIS)
ust 1.2 (Shift-JIS)
If you want to open your file in UTAU later on, try saving as a ust 1.2

I wish I could try this for myself (I just want to see if non-Japanese scripts like hangul work), but unfortunately, it won't even open after installation.
I'm going to have to get together with my Windows-knowing friend to solve that issue, but in the meantime, have you tried running the JAR file directly? Instructions for that on the Github README file, right below the installer links.

Edit: There's no reason why Hangul wouldn't work!

If at all possible, when you start the oto part of develop I highly recommend the option for an extra set of overlap and consonant lines.. (the red and green ones lol) so you can have consonants at the end of samples n stuff... NIAONiao is this way.

Please I beg of you
This is just a frontend made to work with existing UTAU resamplers, so I don't think that'd be possible without making a new resampler too ^^;;
Challenge accepted! I can rewrite the default macres resampler later on to support end consonants. Can't do anything about third-party resamplers though, except maybe something crazy like flipping a sound backwards, running it through the resampler, then flipping it again. :[

Alright, time to pester Kanru! You might be able to reach him yourself on VocalSynth.Space actually.
Actually I left a comment on his bug report page a few months ago with no luck. If you've been in touch with him recently, can you pass on the message?

Double edit: I found and fixed the bug that makes UST files not open in Windows. The fix will be included in my next binary release this weekend. As a workaround, you can open your UST file in Notepad or similar and do a find/replace to replace every instance of "%" with "0.0"
 
Last edited:
  • Like
Reactions: Kiyoteru and Shasta

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
It definitely seems like a lot of this is Windows problems, since I've had no issue with using USTs between UTSU and UTAU-Synth.

EDIT: You can probably get in contact with Kanru more easily via Twitter @khuasw
 

Yoichi-Masaki

Ritsu's Renegades
Defender of Defoko
I did a bit more testing with it, and I noticed a few things with the utau rest notes and utsu's lack of such. This could be part of the issue with the utsu's inability to open usts. I can open a ust made in utsu in utau, but not the other way around, and when I do open an utsu ust in utau, the rests are just longer notes.
untitled_drawing_by_shastaisanandroid-dc08owu.png
untitled_drawing_by_shastaisanandroid-dc08ojw.png


Edit: I use a windows 10 laptop.

Double edit: I'm currently looking at the difference between an utsu ust and an utau ust to see if they're written differently in hopes it'll help. I'll tell you what I find.
I'm on a windows 10 surface pro. It makes sense the rests would be issues, they do the same thing when transferring from utau synth to utau.

Also using other reaamplers has not been working so far.
 

Lethe

Ruko's Ruffians
Supporter
Defender of Defoko
Thread starter
All right, time for a weekend release! You can go here to find new Windows and Mac installers, as well as a JAR executable file that works on Windows, Mac, and Linux.

Updates:
  • Added a playback bar, added bars to mark start and end of highlighted note section.
  • Added auto hiragana/katakana/romaji conversion.
  • Reads prefixmap and auto-adds pitch suffixes for multipitch banks.
  • Auto CV->VCV conversion.
  • Confirmed to work with fresamp, TIPS, and resampler.exe. Does not work with wavtool.exe.
  • Works with moresampler (resampler and wavtool), but you'll need v0.8.4 or higher. (download at https://t.co/VSsFZpp0in)
  • Fixed several bugs that made Windows UTAU files not open in Utsu.
  • Default file type when saving a UST on Windows is now 1.2, not 2.0, for greater compatibility with Windows UTAU.
  • Can highlight multiple notes by shift-clicking. Can un-highlight all by right-clicking the track.
  • Pressing "Render" will only play the notes from the highlighted section. "Export to wav" still exports the entire song every time.
Next plans:
Voicebank editing or plugins. Probably a voicebank editor first. Need to get the core features out before I do anything fancy.

Here's a picture of some of the new stuff:
utsu_screenshot_2.png
 

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
Suggestions
- Q for add notes, W for edit notes, E for erase notes (tool shortcuts in UTAU-Synth)
- Space to play selection
- Rename quantization options to reflect music notation ("1 per beat" -> 1/4, "2 per beat" -> 1/8, and so on)
- When a note is selected, press "enter" to start editing lyrics, and when finished editing lyrics, pressing "enter" will select the following note (like in UTAU-Synth). Alternatively, press "tab" to move to the following note, and still be editing lyrics (like in Vocaloid). It's a bit annoying to have to select each note manually. A method of batch-input for lyrics would be handy too.
- When multiple notes are selected, ability to move them all at once
- Rectangular click+drag selection of multiple notes

I've also tried putting Windows resamplers in a Wineskin wrapper and selecting that as a Mac executable in the project properties. However, no sound is produced. Do you have a better suggested method for using wine for windows resamplers on mac?
 

Nohkara

Pronouns: He/him
Supporter
Defender of Defoko
Hello! May I suggest adding a feature that allows users to edit/fix bad FRQ files. Currently, UTAU-Synth lacks this super important feature and it's unfair that Mac users can't fix these themselves unless they have Windows or running Masao's FRQ editor in WineSkin.
 

Mougeki Mero

Defoko's Slaves
Defender of Defoko
Hello! Tested it and loved so far! I would like to suggest a feature like the use of the command line (?) "$direct=true" from UTAU, which basically plays the note without any pitch correction on them, with the sample only being affected by the oto. I made a tutorial on it once for UTAU:
 

Lethe

Ruko's Ruffians
Supporter
Defender of Defoko
Thread starter
Utsu has a logo now!
Hootsu.png

(Thanks @Soursop the fruit for designing! ^^)

1. Q for add notes, W for edit notes, E for erase notes (tool shortcuts in UTAU-Synth)
2. Space to play selection
3. Rename quantization options to reflect music notation ("1 per beat" -> 1/4, "2 per beat" -> 1/8, and so on)
4. When a note is selected, press "enter" to start editing lyrics, and when finished editing lyrics, pressing "enter" will select the following note (like in UTAU-Synth). Alternatively, press "tab" to move to the following note, and still be editing lyrics (like in Vocaloid). It's a bit annoying to have to select each note manually. A method of batch-input for lyrics would be handy too.
5. When multiple notes are selected, ability to move them all at once
6. Rectangular click+drag selection of multiple notes
1-3: Easy, will do this week
4 & 6: Medium difficulty, will do soon-ish
5. Hard, will do eventually

Edit: I haven't played around with Wine yet. Could you get the wrapped resamplers to work by giving them test data in the command line? If they work on the command line, they can work on Utsu.

May I suggest adding a feature that allows users to edit/fix bad FRQ files. Currently, UTAU-Synth lacks this super important feature and it's unfair that Mac users can't fix these themselves unless they have Windows or running Masao's FRQ editor in WineSkin.
Sure, I'll do that as part of the voicebank editor. ^^

I would like to suggest a feature like the use of the command line (?) "$direct=true" from UTAU, which basically plays the note without any pitch correction on them, with the sample only being affected by the oto.
So it plays a note without any pitchbends? I could add that as a UST or a voicebank setting maybe

If there is something I'm curious of, is this program built for Windows 10 and above?
I built and tested it on Windows 10. Nobody has tested it on Windows 8 or lower, though.
 
Last edited:

Mougeki Mero

Defoko's Slaves
Defender of Defoko
So it plays a note without any pitchbends? I could add that as a UST or a voicebank setting maybe
I liked the logo, what was the concept behind it?

Basically...Well, actually "$direct=true" remove any pitch correction, not only pitchbends, but the "auto tuning" thing UTAU does. Im really sorry for the bad explanation. It is more or less like this:

Regular use of UTAU: When you record a sample at G4, and play it as a D4 note inside UTAU, the sample will be corrected and plays a D4 note.

$direct=true: When you play a G4 recorded Sample as a D4 note inside UTAU, it will play as a G4 sample always, regardless of the note it was placed.

Althought for voicebanks it doesnt make much difference, for instrument banks it is a huge difference. My acoustic guitar bank, for example, when used regular in UTAU will have problems playing on higher notes like C5+ because UTAU tries to correct the sample's volume, pitch and frequency to match a voice. If I use this same bank using "$direct=true" it removes any correction UTAU does and play the raw sample, only being affected by lenght of the note, thus giving a much more realistic sound. Here is a comparison I made with my piano UTAU: https://clyp.it/v3zdwtzr

Im sorry for the bad explanation, but hopefully I did it correct this time.
 

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
@Soursop the fruit
It's an interesting logo, but seems something more suited for a billboard than a small application icon. You'll probably want to simplify it.
1-3: Easy, will do this week
4 & 6: Medium difficulty, will do soon-ish
5. Hard, will do eventually
Thanks for listening to my suggestions!
Could you get the wrapped resamplers to work by giving them test data in the command line? If they work on the command line, they can work on Utsu.
How exactly do I go about doing this?
 

Lethe

Ruko's Ruffians
Supporter
Defender of Defoko
Thread starter
If I use this same bank using "$direct=true" it removes any correction UTAU does and play the raw sample, only being affected by length of the note, thus giving a much more realistic sound
One more question: When "$direct=true" is set, does UTAU also ignore the consonant length, preutterance, and overlap from the OTO file? If so, I'll be able to implement it without much trouble.

@Kiyoteru
First, open the command line.
I'll use macres as an example. You can find macres executables for mac, windows, and linux here.
To run the Mac executable through the command line, navigate to the same directory and type:
> ./macres
The resampler should spit out help text.

Running a Windows executable through the command line is harder. If you have wine (the command-line tool) installed, you can navigate to the same directory as the Windows executable and type:
> wine macres.exe
If wine is working right, the resampler will spit out the same help text as before.

If you use Wineskin, I believe the final output is a .app file. To call a .app file from the command line, first make sure the .app file is in your /Applications directory. Then you can type:
> open -a Macres.app
("open -a Macres" also works)
Again, this will print out the macres help text if everything is working right.

@Khento
Thanks! Utsu has no planned web version, so hopefully I won't be stepping on your toes with Melodia Synth.

---
I'm still working on the voicebank editor. It's in a gruesome half-finished state right now, but here's teaser picture anyway:

Screen Shot 2018-01-30 at 11.03.43 PM.png

View attachment 6865
 
Last edited:

Kiyoteru

UtaForum power user
Supporter
Defender of Defoko
Running the Mac executable worked fine as expected.
Running the Windows executable through Wineskin didn't work at all. It tried to open the app wrapper, and didn't output anything to the terminal window.
I kicked my computer around a bit and managed to get regular old wine to work. Running the Windows executable using wine worked, but it took a long time. It gave me two timeout error messages before finally outputting the same help text as the Mac executable.
Code:
:err:process:__wine_kernel_init boot event wait timed out
 

lunari162

Defoko's Slaves
Defender of Defoko
One more question: When "$direct=true" is set, does UTAU also ignore the consonant length, preutterance, and overlap from the OTO file? If so, I'll be able to implement it without much trouble.
I can't speak for UTAU-Synth, but on PC, I'm 99% sure it doesn't ignore the oto settings. If you use $direct=true on a VCV note, it'll still use whatever preutterance and overlap you set. It's almost like those settings tell it how much to "directly play." I'll double check it later if I have the time!
 

Similar threads