AckieSound, creator of VocalShifter, VS4U, and other software, has recently announced their UTAU VST plugin project.
Download: http://ackiesound.ifdef.jp/soko.html#beta
English Manual: https://tl.tubs.wtf/2021/08/15/parallel-world
Real-time singing synthesis in a DAW using UTAU voicebanks.
Since vocals are being synthesized in real time, all changes will be reflected in real time, whether you change the lyrics, timing, or the entire UST (as long as new frequency files don't need to be generated).
For the purpose of this VST plugin, a dedicated real time singing synthesis engine was developed. It was created using WORLD to combine the functions of the resampler and wavtool. (WORLD can overlap audio without taking phase alignment into account, so it's relatively easier to do real time synthesis.)
The UTAU editor runs independently of the VST plugin, and is used to edit UST files. (It doesn't actually show up as part of the VST plugin window.)
The VST plugin detects when a UST file has been updated and rereads the file, and the plugin by itself synthesizes the vocals.
If you have a voicebank or UST that doesn't work normally in Parallel World, please send information to AckieSound. If there are alarm beeps or if the plugin itself doesn't work, try using the engine test tool. This will show the cause of the error.
Download: http://ackiesound.ifdef.jp/soko.html#beta
English Manual: https://tl.tubs.wtf/2021/08/15/parallel-world
Real-time singing synthesis in a DAW using UTAU voicebanks.
Since vocals are being synthesized in real time, all changes will be reflected in real time, whether you change the lyrics, timing, or the entire UST (as long as new frequency files don't need to be generated).
For the purpose of this VST plugin, a dedicated real time singing synthesis engine was developed. It was created using WORLD to combine the functions of the resampler and wavtool. (WORLD can overlap audio without taking phase alignment into account, so it's relatively easier to do real time synthesis.)
The UTAU editor runs independently of the VST plugin, and is used to edit UST files. (It doesn't actually show up as part of the VST plugin window.)
The VST plugin detects when a UST file has been updated and rereads the file, and the plugin by itself synthesizes the vocals.
Past updates
Now undergoing closed beta testing. Those who would like to participate should inquire via Twitter DM. The current version is unstable and has many limitations, so for those who are unsure we recommend waiting until a stable version has been released.
To all Parallel World testers:
ver. 0.02 has been released. A bug has been fixed where there was no audio output when a full path was specified for VoiceDir.
To all Parallel World testers:
ver 0.03 has been released. A tool has been created to check the function of the engine without using a DAW. Real time reflection of changing voicebank is now supported.
To all Parallel World testers:
ver 0.04 has been released. BRE and g flags are now supported. An error now occurs when a voicebank folder is under Program Files. A feature was added to indicate errors by playing a sound.
To all Parallel World testers:
ver 0.05 has been released. The t flag is now supported. A bug where BRE wasn't applied normally has been fixed. W flag (growl) and S flag (devoicign) have been added.
Ver. 0.05's currently supported features.
Midnote flag changes aren't in the UST specifications, and there's no way to modify them in the editor, so there's no plan to support them. It's included simply because the engine supports it. For expression voicebanks, if the path to the voicebank to be morphed could be specified somehow, a flag could be used to specify the ratio between the two banks.
To all Parallel World testers:
ver. 0.06 has been released. Tempo changes are now supported. A bug has been fixed where Rendering Options in the project settings were not applied.
To all Parallel World testers:
ver. 0.07 has been released. Multiple channels are now supported. Multiple UST files can be assigned to separate tracks, and vocals will all be synthesized at the same time.
The remaining features are automatic CVVC conversion, Unicode support, and Mode 2 rendering. Automatic CVVC conversion and Unicode support will be delayed a little, so if Mode 2 can be supported then I'm thinking of moving them to the open beta test.
To all Parallel World testers:
ver 0.08 has been released. Mode 2 is now supported. The highest point in the portamento determines which pitch in the prefix.map is used. Error logging has been added.
Because the lowest point of the pitchbend determines which original audio is used, it now prevents the audio quality deteriorating when the pitchbend goes lower than the note in the UST.
To all Parallel World testers:
ver. 0.10 has been released. Switching between Mode 1 and Mode 2 is now supported. This can be changed in UTAU in Project > Project Property > Rendering Options. Type M1 to use Mode 1 and M2 to use Mode 2.
Parallel World is now available to the general public. Synthesize vocals with UTAU in your DAW with real-time rendering.
Parallel World ver. 0.12 has been released. Added support for morphing with expressive voicebanks. Fixed 2 bugs.
Bugfixes:
Crash when using a nonexistent alias.
When specific notes are used together, memory consumption continued increasing during playback.
Example of the morphing function. The vocal tone here changes in the order of "Yamine Renri Ver1.0" to "Yamine Renri WHISPER" to "Yamine Renri CLEAR"
Parallel World ver. 0.13 has been released. Fixed bug where oto.ini wasn't read normally when alias isn't specified.
The %VOICE% setting in Parallel world should not be one individual voicebank folder, but the folder for all voicebanks. If you have not changed your UTAU settings, it should be the folder named "voice" inside of UTAU's instatllation folder. (This setting is actually used to substitute %VOICE% of VoiceDir in the UST.)
Parallel World ver. 0.14 has been released. Fixed a crash when an invalid voicebank was used. Support for drag and drop of UST files into settings.
To all Parallel World testers:
ver. 0.02 has been released. A bug has been fixed where there was no audio output when a full path was specified for VoiceDir.
To all Parallel World testers:
ver 0.03 has been released. A tool has been created to check the function of the engine without using a DAW. Real time reflection of changing voicebank is now supported.
To all Parallel World testers:
ver 0.04 has been released. BRE and g flags are now supported. An error now occurs when a voicebank folder is under Program Files. A feature was added to indicate errors by playing a sound.
To all Parallel World testers:
ver 0.05 has been released. The t flag is now supported. A bug where BRE wasn't applied normally has been fixed. W flag (growl) and S flag (devoicign) have been added.
Ver. 0.05's currently supported features.
Supported | Supported but unusable | Unsupported |
---|---|---|
Unlimited # of aliases | Midnote flag change | Automatic CVVC conversion |
multi-prefix | Expression voicebank morphing | Unicode support |
Generate frequency files in parallel | Mixing multiple voice tracks | Multi channel support |
Automatic VCV conversion | Mixing multiple audio tracks | Tempo changes |
Real time singing synthesis | Mode 2 pitchbends | |
BRE, g, and t flag | ||
Growl and devoicing | ||
Mode 1 pitchbends |
Midnote flag changes aren't in the UST specifications, and there's no way to modify them in the editor, so there's no plan to support them. It's included simply because the engine supports it. For expression voicebanks, if the path to the voicebank to be morphed could be specified somehow, a flag could be used to specify the ratio between the two banks.
To all Parallel World testers:
ver. 0.06 has been released. Tempo changes are now supported. A bug has been fixed where Rendering Options in the project settings were not applied.
To all Parallel World testers:
ver. 0.07 has been released. Multiple channels are now supported. Multiple UST files can be assigned to separate tracks, and vocals will all be synthesized at the same time.
The remaining features are automatic CVVC conversion, Unicode support, and Mode 2 rendering. Automatic CVVC conversion and Unicode support will be delayed a little, so if Mode 2 can be supported then I'm thinking of moving them to the open beta test.
To all Parallel World testers:
ver 0.08 has been released. Mode 2 is now supported. The highest point in the portamento determines which pitch in the prefix.map is used. Error logging has been added.
Because the lowest point of the pitchbend determines which original audio is used, it now prevents the audio quality deteriorating when the pitchbend goes lower than the note in the UST.
To all Parallel World testers:
ver. 0.10 has been released. Switching between Mode 1 and Mode 2 is now supported. This can be changed in UTAU in Project > Project Property > Rendering Options. Type M1 to use Mode 1 and M2 to use Mode 2.
Parallel World is now available to the general public. Synthesize vocals with UTAU in your DAW with real-time rendering.
Parallel World ver. 0.12 has been released. Added support for morphing with expressive voicebanks. Fixed 2 bugs.
Bugfixes:
Crash when using a nonexistent alias.
When specific notes are used together, memory consumption continued increasing during playback.
Example of the morphing function. The vocal tone here changes in the order of "Yamine Renri Ver1.0" to "Yamine Renri WHISPER" to "Yamine Renri CLEAR"
Parallel World ver. 0.13 has been released. Fixed bug where oto.ini wasn't read normally when alias isn't specified.
The %VOICE% setting in Parallel world should not be one individual voicebank folder, but the folder for all voicebanks. If you have not changed your UTAU settings, it should be the folder named "voice" inside of UTAU's instatllation folder. (This setting is actually used to substitute %VOICE% of VoiceDir in the UST.)
Parallel World ver. 0.14 has been released. Fixed a crash when an invalid voicebank was used. Support for drag and drop of UST files into settings.
Latest update
Parallel World ver. 0.15 has been released. Fixed a crash when full width characters are used in flags. Improved behavior when using invalid voicebank settings. Increased the maximum number of characters in lyrics. Enhanced error checks.If you have a voicebank or UST that doesn't work normally in Parallel World, please send information to AckieSound. If there are alarm beeps or if the plugin itself doesn't work, try using the engine test tool. This will show the cause of the error.
Last edited: