VOCALOID4 (commonly shortened to just "V4") was the successor to the VOCALOID3 engine by the YAMAHA Corporation.
VOCALOID4 was first mentioned by Bil Bryant of PowerFX Systems AB. during a discussion about Ruby, a then unreleased English VOCALOID. The new engine's existence was made known in late October 2014.
On November 24, Internet Co., Ltd., AH-Software Co. Ltd. and Crypton Future Media, Inc. spoke about the development direction they were taking for the new engine. For AH-Software, due to the impact Windows 10 had on maintaining the VOCALOID2 software, AH Software decided to relaunch every VOCALOID2 vocal they had produced as a VOCALOID4 vocal. Due to the fact that the provider of Kaai Yuki would likely be of middle or highschool age, re-sampling her voice for growl samples was a problem. It was decided to find a second voice provider for the purpose. Internet co noted that they would try to increase triphone numbers and were working to increase the overall performance of their vocals. Internet co., Ltd also confirmed that they would release a new vocal, so their direction for VOCALOID4 was also focused more on updating old vocals rather than creating new ones.
The engine worked with VOCALOID NET, though this service was later retired on 27 September, 2016.
On January 30, 2015, Kenmochi Hideki retired from Vocaloid development, making VOCALOID4 the last version of the VOCALOID engine he worked upon. He was replaced with Katsumi Ishikawa.
On February 20, a special tool was released which allowed VOCALOID2 vocals to be imported into VOCALOID4 without the need of VOCALOID3.
In April 2015, a iOS version was released called Mobile VOCALOID Editor.
Impact on voicebank development
RUBY was previously planned to be released around Christmas 2014, but was delayed since V4 was due for release "soon".
Updated versions can be downloaded on the official Download page.
- Ver.4.0.1: was released on December 15, 2014 and Lily was added to the Cross-Synthesis Group, along with general improvements to stability.
- Ver.4.1.0: was released on March 3, 2015 and support for Luka V4X was added, as well as the ability to revert selected notes to default lyrics, enter multiple lyrics for a single note, and general improvements to stability.
- Ver.4.2.0: was released on August 5, 2015 with the option to activate unactive voicebanks from start up, Selection History and the Swap Buttons for Control Parameters, the listing a Job Plugin as a "Job" Menu Item, the combination of three or more parts of VOCALOID Tracks, along with general improvements to stability.
- Ver.4.2.1: was released on September 10, 2015, featuring only general improvements to stability.
- Ver.4.3.0: was released on October 25, 2016, featuring an updated "XSY" group, removing of the VOCALOID.net feature (due to its closure), and general improvements to stability. The general VOCALOID API had been updated to allow this with this version, this also impacted Piapro Studio allowing it to also use this function in its own version 188.8.131.52 update.
- OS: Windows 10, Windows 8.1, Windows 8, Windows 7 SP1 (32/64bit)
- CPU: Intel Dual Core CPU
- RAM: 2GB or more
- Recommended: 4GB or more
- HDD: 500MB free space or more
Note these requirements only apply to the VOCALOID4 engine itself. Users need to check each VOCALOID package and its requirements before installation as some variation may occur between releases.
Also note that if you own VOCALOID4 Editor for Cubase, you do not also have to purchase the full VOCALOID4 software.
Examples of usage
For a list of VOCALOID4s new parameters see Parameters
The new engine allows the importation of VOCALOID2 and VOCALOID3 vocals; initially VOCALOID2 vocals had to be imported into VOCALOID3 before they could be imported into VOCALOID4. Yamaha has since released a tool to by-pass this. The new engine includes other features, but not all of them are accessible to VOCALOID2 and VOCALOID3 vocals. The new version of the engine includes a "Growl" (GWL) feature which allows vocals built for VOCALOID4 to take on a growl-like property in their singing results.
Cross-synthesis (XSY) was also added, allowing the user to switch between two vocals of the same language and voice (vocals from the same VOCALOID) smoothly using a time-varying parameter. It originally would not work with vocals from different characters, even if they are held within the same package. For example, using this feature with anon & kanon was not possible. This feature is accessible to VOCALOID3 and VOCALOID4 vocals, but not VOCALOID2. Later, the feature was expanded and "XSY groups" were introduced. Vocaloid within this group could have their vocals mixed together.
Another feature included is Pitch Rendering, which all imported vocals can use. This displays the effective pitch curve on the UI. Finally, real-time input has been included in this version; a feature which all vocals can use. Other features were also noted by DTM magazine in their follow up page. Among them, are the 'Convert Vibrato to Pitch Bend' and 'Convert Expression to Pitch Bend' options, which respectively converts the vibrato and legatos/portamento normally edited through the Note Expression Property window into equivalent curves of the PIT parameter. Another new option is the 'Master Tune', which sets the overall tuning of the VOCALOID Editor. The default value is 440Hz for the note A, and modifying this value will change the respective pitch and associated frequency for all the notes.
For users who had bought VOCALOID3, VOCALOID for cubase or VY1v3, YAMAHA offered an upgrade for each software package at the price of 5400 JPY. Those who bought the editor after November 10th 2014 were also offered a free upgrade until June 2015.
It has been confirmed that VOCALOID4 vocals have a larger number of triphonetic samples.
As of CYBER DIVA, issues with English VOCALOIDs have been addressed and a new script has been created for recording use to develop them. The script gives more clarity to English vocals, however, some expressive tones are lost in the process. It also addresses pronunciation issues. In addition, it was confirmed that past English vocals suffered from mis-labeled sounds; the new script helps reduce the errors.
However, this new script for English was not shared as the "standard" script as soon as it was written, with RUBY's script being made from scratch by Syo, as the previous YOHIOloid script contained errors and lacked efficiency.
V4 also saw a change in approach to Vocaloid recordings. In an interview for Unity-Chan for the later VOCALOID4 engine, "Baba" (吉田) reported that for his series with this engine a lot of the focus turned to bringing out the qualities of the vocalist. Until V3, the vocals were fixed on producing results that easily fit music as a result of various common practices by he and his fellow developers. The result was, there was not always as great of a difference between vocals as they were for the vocals he worked on for Vocaloid4. The process involved examine the character and what traits needed to be brought out, such as more vivid tones for a cheerful and bright character. The new recording technique was developed by Internet co., Ltd and was first demonstrated with Chika, a VOCALOID3 vocal.
While it is possible to use VSQX files from VOCALOID3 in VOCALOID4, the files saved in VOCALOID4 cannot be used in VOCALOID3.
Despite the new recording style and the introduction of more triphones for many Vocaloids, there are no major developments for Japanese Voicebanks. This is in contrast to the leap from VOCALOID2 to VOCALOID3 due to the addition of triphones, which had greatly improve the engine overall. The biggest changes in this engine were to English voicebanks, many of which saw overhauls, though not all developments were universally shared by all. There is no information on the situation with other languages. For Spanish vocals, this is simply because no Spanish voicebanks were produced for VOCALOID4, thus no changes were made in VOCALOID4 as a result.
VOCALOID4 has issues when it comes to detecting problematic phonemes and can impact phonetic complex languages like English.
In comparison to VOCALOID3 the amount of focus on the introduction of the newer engine was far less. There was no large event to announce it, and it was released within a month of it being leaked via PowerFX. Though studios knew months in advance like past versions, VOCALOIDs like Chika and Ryuto saw releases right until VOCALOID3s final weeks with no large release gaps like there had been between VOCALOID2s end and VOCALOID3s release. In addition the Chinese version of the software did not receive an update at the same time as the English and Japanese version, meaning Chinese VOCALOID3 releases were still being made after VOCALOID4 was released. In addition the initial release of this engine only focused on English and Japanese, with the other 3 languages lacking support.
DTM MAGAZINE at this point ran "VOCALOID NEWS" with every issue of its publication and the magazine at times focused more and more on VOCALOID right until its final issue in 2016.
In addition other changes were made with the handling of the softwares support overall. VOCALOID STORE was replaced with the VOCALOID SHOP, and physical copies were not offered for many voicebanks it sold. 3rd party studios such as AH-Software Co. Ltd. and Internet Co., Ltd. offered physical copies via their own websites, but many VOCALOID had become "download only". VOCALOID accounts were still active . The VOCALOID NET was closed, with the cloud service being removed during updating of the VOCALOID4 software itself.
Though there were differences in capabilities between V3 and V4, these were often notable not to be as significant as the differences between V2 and V3.
This wasn't just restricted to the engine either, but to the Vocaloid released for it. For example, Rana's V3 vocal was the same as the V4 vocal just with the addition of GWL capability added. As a result Rana V3 imported into V4 produced the same results as Rana V4 if the GWL feature wasn't used. Tone Rion V4 is yet another example of such a release and she is the same for her VOCALOID4 release as her original V3 release, though in addition to GWL XSY was added to allow her to XSY with Yumemi Nemu.
VOCALOID4 had more releases then VOCALOID3 overall, but many of them were just re-releases of older VOCALOID2 or VOCALOID3 products, over half that of VOCALOID3. There were overall more non-Japanese newcomers compared to Japanese a marked difference to VOCALOID3 where less then half of the new comer vocalists were non-Japanese ones. There were no Spanish voicebanks released at all for the engine.
The VOCALOID4 engine was generally stable and was a suitable addition to a musician studio, though still unable to fully replace real singers.
Between the closure of NETVOCALOID, the retirement of DTM magazine and the departure of a number of several developing studios by the end of VOCALOID4, VOCALOID4 was considered far more mediocre in reception towards the end then the beginning. The general feeling of the end of the VOCALOID craze, however, would be more felt in VOCALOID5.