Vocaloid Wiki
Advertisement
🛠 This subject is work in progress.
Please bear with us while improvements are being made and assume good faith until the edits are complete.
For information on how to help, see the guidelines. More subjects categorized here.
🛠
📰 This subject requires intervention. What is being worked on: Please note the details of this software are subject to change without notice.
For information on how to help, see the guidelines.  More subjects categorized here.
📰

VX-β is an upcoming VST3/AU plug-in voice synthesis engine to be bundled soon with the VOCALOID6 editor. A prototype version of this plug-in was originally included within the VOCALOID β-STUDIO research studio developed by the company Yamaha Corporation. It functions within various digital audio workstation (DAW) applications. VX-β is equipped with AI-based deep learning technology that uses singing voice synthesis technology derived from the research or "beta" stage that VOCALOID β-STUDIO represented.

The prototype version of the software was released via lottery application to eligible applicants beginning August 23, 2023;[1] the final version, announced on July 18, 2024, is expected to be released at a future, yet-undetermined date.[2]

History[]

VOCALOID β-STUDIO was first teased on July 27, 2023, featuring a promotional image of the "β" (beta) Greek letter and a date of August 8, 2023 with no additional details provided.[3] Yamaha officially announced the project on August 8, revealed as a limited research studio to be made available to creators with the hope that they would use the rough prototype software provided that was still in the research stage and has unknown value, to create works that would open up new horizons in the field of singing voice synthesis. The first project, the VX-β digital audio workstation (DAW) plug-in, was simultaneously announced the same day. It was announced that the VOCALOID β-STUDIO services would have further details released in late August. The date of services to end was announced to be March 31, 2024 and the VX-β DAW plugin service was announced to end on April 1, 2024.[4]

On August 22, 2023, Yamaha began accepting applications for participation in the VOCALOID β-STUDIO campaign and encouraged the creation of new music using the VX-β application. It was noted that in the future, Yamaha would recruit new applicants irregularly, and in addition to those who have won the lottery, the winning applicants who have been selected based on the answers to the questionnaire would receive the VX-β DAW plug-in, which uses singing voice synthesis technology that was still in the research stage, and would be distributed free of charge.[5][6] Music created using VX-β was subject to the individual character terms of use of each voicebank.[7] Users could freely publish within the scope of the terms. A total of nine voicebanks were announced for beta testing of the software including: prtv_0, prtv_1, prtv_2, prtv_3, Gazenβ, nagiβ, multiβ-N (which includes an additional seventeen individual voicebanks within the main multiβ-N voicebank package), Gekiyakuβ and Kazehikiβ.

Beginning August 23, the first round of recruitment began and winning applicants were presented with a download and serial codes for the VX-β plugin. Those who were not selected at the time would automatically be enrolled for the next recruitment.[8]

Development[]

The first project of the VOCALOID β-STUDIO was announced to be the release of VX-β, a DAW plug-in based on Yamaha's research-stage voice synthesis technology. It was noted that VX-β was not considered a beta version of the next version of VOCALOID products. It was an experimental product that incorporated functions and elements that proposed a more futuristic world of voice synthesis. Yamaha's hope was that creators would challenge new singing voice synthesis with their own free ideas and make use of this product for their own creations. VOCALOID β-STUDIO would be available for all creators to participate. Random drawings were announced among those who applied for participation and serial codes to use VX-β sequentially would be issued to applicants. While Yamaha continued to enhance the project structure in order to deliver VX-β to as many people as possible, delivery times for new products would be limited. It was noted that after April 1, 2024, VX-β would no longer be available for activation due to the end of the usage period. Details on the application for participation would be available in late August 2023. Services offered by VOCALOID β-STUDIO would be limited to March 31, 2024.

Lottery Application[]

Applicants could apply to participate in VOCALOID β-STUDIO from the Yamaha Music Members campaign page. To apply, applicants had to (1) obtain a Yamaha Music ID and (2) answer a questionnaire.

A random lottery was held for those who applied during the period (until March 31, 2024), and the serial code and download link of the singing voice synthesis plug-in VX-β were sent to the winners by email. It was noted that some of the winners were selected based on the answers to the questionnaire, in order to create many works that expand the possibilities of new singing voice synthesis with VX-β.

About the Lottery Procedure[]

The number and schedule of irregular lottery procedures and the number of winners have not been disclosed. Applicants could check the lottery status after applying from the "Application History" on the Yamaha Music Members website. Those who were not selected in one lottery would be eligible for the next lottery. The same person could not win the lottery more than once. If applicants would like to modify their answers to the questionnaire after completing the application, applicants were allowed to complete the application procedure again. If applicants applied multiple times, only the last one would be valid. Note that Yamaha cannot answer questions about the lottery process and content.

Final release[]

On July 18, 2024, VOCALOID β-STUDIO posted its very last tweet, announcing that the account would be closing soon and thanking everyone who participated in the program. It also stated that new announcements regarding the program would be made instead on the official Yamaha VOCALOID Twitter account. The same tweet ended with an embedded video announcing that the VX-β plug-in would be in the near future bundled into the VOCALOID6 editor and made available for use by all VOCALOID6 users, requiring only the same serial code used to authorize VOCALOID6 and its standard voicebanks (now called the Silhouette Series); it was stated that all AI voicebanks (both current and future) would be made available for VX-β as well. It also announced that two of the debut VX-β vocals, Gekiyaku and Kazehiki, were now available for VOCALOID6 and that the last four VX-β vocals (Ci-chan, Kanade Kanon, And Uge, and Kasukabe Tsumugi) would be in the future be released for both VOCALOID6 and the upcoming VX-β plug-in.[9][2][10]

Requirements[]

  • OS: Windows 10, Windows 11, macOS 13 Ventura
  • CPU: Haswell (4th generation) or later Intel Core series or Xeon series; Apple Silicon
  • RAM: 8 GB minimum recommended
  • HDD: At least 1 GB or more (including all voicebanks)
  • Graphic Minimum Open GL 3.2 or higher
  • Other: Audio device, internet connection (for authentication, deauthentication, software updates, etc.)
  • Monitor size:
    • Minimum operating environment: 1024×768 or higher resolution

Additional installation notes[]

Works created with VX-β can be freely published and can be used commercially. However, it was noted that participants would need to comply with the VX-β terms of use when publishing songs, and also comply with the rules of each individual voicebank character.

Usable Period[]

The prototype version of VX-β had a limited usable period. After March 31, 2024, the activity end date of this project, it will not start as a plug-in and cannot be used. It was encouraged that users save important work still being worked on by exporting it onto their respective DAW software until then.

The final version of VX-β, which will only require a valid VOCALOID6 serial code for authorization, will have no usage time limits.[9]

Releases[]

VOCALOID β-STUDIO logo blk

Vocal libraries released for the VOCALOID β-STUDIO VX-β engine.

Announced Vocals[]

Announced Vocals for the VX-β software. Names and images presented are placeholders and may not reflect the final product when complete.

We urge readers be considerate and careful when supporting unconfirmed projects, especially if they ask for financial endorsement.

Additional notes[]

Examples of usage[]

An example of solfège using VX-β technology. See also, a listing of vocal stats here.

To contribute an example- see this blog entry to download the VSQx.

prtv_0
prtv_1
prtv_2
prtv_3
Gazenβ
nagiβ
multiβ-N:
f00
f01
f02
f03
f04
f05
f06
f07
f08
f09
f10
f11
m00
m01
m02
m03
m04
Character vocals:
Gekiyakuβ
File:Gekiyakuβ - Solfege.ogg
Kazehikiβ
File:Kazehikiβ - Solfege.ogg
Kanade Kanon β
File:Kanade Kanon β - Solfege.ogg
And Uge β
File:And Uge β - Solfege.ogg
Ci-chan β
File:Ci-chan β - Solfege.ogg
Kasukabe Tsumugi β
File:Kasukabe Tsumugi β - Solfege.ogg

New features[]

For a list of VX-βs new parameters see Parameters

VX-β offers new features including:

  • Voice: VX-β does not require downloading of individual voicebanks, as all voicebank data was included in the plug-in.
    • Note that voicebank data may be added or changed in future version upgrades.
  • Singer/Style: Typically, one voicebank represents one singer. This pull-down switches the singing style that was unique to the singer represented by a given voicebank. Switching between singing styles switches not only the pitch movement, but also the timbre of the singing voice. Selecting a voicebank would take time off of the sequence length, but switching between singing styles would be instantaneous. The exception was the voicebank multiβ-N, which contained multiple singers within it. multiβ-N uses this pulldown to switch the singer itself, not the singing style. f00 to f11 are female singers and m00 to m04 are male singers. Users are able to use the new editing tools to freely edit the accents, vibrato, rhythmic feel, and more.
  • Various Parameters: By manipulating various parameters, users can increase the expressiveness of the sound. Each parameter can be intuitively manipulated and followed in real time during playback. Any of the parameters can be changed by dragging with the mouse, entering numerical values by clicking on the numerical value section, or restoring default values by double clicking on the knob or slider.
    • Parameters include: Air, Formant, Attack, Vibrato, Power, Timing, Pitch, Fuzzy, Presence, Kero, Key, Master, Tune, & Output
  • Use of Automation: All parameters can be dynamically controlled using the DAW's automation functions.
  • Keep Voicing, Guide Tone, Visual Functions:
    • Keep Voicing Function: When the Keep Voicing function was turned on, the sound of the song position position can be continued when playback was stopped. This function allows users to make adjustments while checking the changes in voice tone and strength/weakness caused by each parameter in real time.
    • Guide Tone Function: A button that toggles on/off the guide sound for note input.
    • Visual Functions: The waveform can be displayed or hidden. By drawing a waveform of a singing voice being synthesized in real time, the pitch and volume can be visually captured. Turning it off reduces the drawing load.
  • Other Functions:
    • Breath Insertion and Cutting: If there was a gap between notes, it may be considered a rest and a breath may be automatically inserted. Similarly, when a note was separated, a breath was inserted, so if users do not wish to include a breath, users can input the lyric "っ" and it would be pronounced as a natural prompt. If users wish to force the insertion of a breath, users can do so by entering "BR" and the diacritical marks as shown in the list of special diacritical marks below.
    • Vowel Voicelessness: You can silence vowels by adding a "0" (half-width zero) to the lyrics. For target notes that users wish to silence, such as "de shi 0 ta" or "so shi 0 te," add a "0" to the lyrics.
  • Phonetic Symbols:
    • Basic Information on Pronunciation Symbols: VX-β does not allow the phonetic symbols for VOCALOID to be input as they are. When loading a .vpr file (VOCALOID5/6 sequence), the diacritics for VOCALOID written in the .vpr file are automatically converted to diacritics for VX-β. If you want to change the automatically converted diacritical marks, you can use the VOCALOID5/6 Editor's diacritical marks field for VX-beta You can directly write the diacritics for VX-β by prefixing them with a $. For example, To pronounce "fu" = [f u], enter $f $u in the VOCALOID5/6 diacritical marks field.


Gallery[]

Media Gallery[]


Tutorials[]


The reference manual for this program can be found here.

References[]

Navigation[]

Advertisement