ROSA

This project is currently in production.

According to official sources, this project was announced to currently be in production or is in its conceptual stages.
Please read the page for more information.

ROSA	SynthV Studio	CeVIO AI	|	Originals	|	Gallery

ROSA
Character Information
Gender	Female
Age	15
Birthday	December 17
Height	152 cm (5 ft)
General Voice Database Information
Voice	Feminine
Voice Provider	Shirayuki
Base Language(s)	Japanese
Additional Product Information
Illustrator(s)	Yoshida Iyo
Company	ZAN-SHIN Co., Ltd.
Release Date(s)	June 2024
Status	In development
Affiliations	Internet Co., Ltd. Dreamtonics Co., Ltd. AH-Software Co. Ltd.
Website	ZAN-SHIN Co., Ltd. Official ROSA Website
Examples of Usage
List of Songs	VocaDB Voice Synthesizer Lyrics Wiki

ROSA (ロサ) is an upcoming Japanese vocal to be developed by ZAN-SHIN Co., Ltd. in collaboration with AH-Software Co. Ltd. and distributed by Internet Co., Ltd., and is expected to release as an AI voice database for Synthesizer V Studio. She was previously released as a CeVIO AI speech voicebank in December 2021. There was an interest in the possibility of creating singing and speaking voice libraries in English for both engines.

Her voice provider is Shirayuki, a Japanese influencer and voice actress. She was the chosen candidate from ROSA's audition.

Concept

ROSA's theme color is pink and she is the youngest sister of the VOCALOID3 CUL. She is a cheerful and mischievous person.

ROSA was born in the Shinobi No Sato Ninja Village (忍びの里) with her older sisters which was under constant threat by hostile forces and as such, was separated from her sisters for a long period of time. Unbeknownst to her, she was entrusted to an old little church on a small hill where she lived together with the nuns to keep her safe. ROSA, who was actually the heir to CUL, had her name changed so that her relationship with her sisters was unknown. For much of her life, she had believed she was alone and often questioned who she was and what her purpose in life is. One day CUL appeared to her and told her that she was ROSA's older sister.

Etymology

"ROSA", a Spanish word, can mean "rose" (as in the flower), but was selected mainly for its relation to the color pink (which in languages like German and Spanish is also called rosa).

Appearance

ROSA wears a robe similar to a traditional Japanese Kimono, consisting of a black and pink top and a short pleated maroon colored skirt. The sleeves are decorated with soft pink roses and a curly white pattern. Around her waist is a black Obi (traditional Japanese belt worn around the waist.) laced together in the front. A red bow is tied in the back. The skirt of her top is adorned with the same wavy pattern as the sleeves. Her shoes are tall black lace-up boots reaching her knee. The heels are a deep red. She wears a black choker with silver detailing. Her hair is cut at an angle, stylized in a high braid to one side of her head. The color is a cotton candy pink with a crimson intertwined in the braid. She has many hair accessories, such as a red bow, rose hair pins, and a black ribbon. ROSA has reddish-pink eyes, and rosy cheeks.

Relations

• WINDCUL (青姉 Aonee "Blue Sister") - older blue and wind-themed sister (19 years old; conceptual)
• WhiteCUL (雪さん Yuki-san) - older white and snow-themed sister (20)
• MOONCUL (つき Tsuki) - older yellow and moon-themed sister (17; conceptual)
• CUL (also known as リカル RE:CUL or REDCUL) - older red and flower-themed sister and BLACKCUL's twin (16)
• BLACKCUL (黒CUL Kuro CUL, a.k.a. 濡鴉 Nuregarasu "Wet Crow") - older black-themed sister and CUL's twin (16)

History

2021

On March 20, 2021, HirotoP confirmed that ROSA, CUL's younger 15-year-old sister with a pink theme, was in production for a new AI singing and talk synthesizer developed by ZAN-SHIN Co., Ltd., seeming to suggest that VOCALO Revolution was not producing vocals for the VOCALOID software.

It was mentioned that ROSA's voice actress would be chosen via an audition taking place in April. HirotoP also discussed the time when Eri Kitamura recorded for CUL and that she was directed to make a voice that was completely different to other VOCALOIDs, which resulted in a strong voice. The voice direction for ROSA's actress would be treated the same, though it depended on the results of the audition. He expressed an interest in creating ROSA's singing voice database for Synthesizer V AI or CeVIO AI and a speech voicebank for A.I.VOICE or CeVIO AI, but the final decision would not be announced until April. In addition, due to an unexpectedly high interest from western fans, HirotoP was also considering possible English singing and speech voicebanks for ROSA.

On March 26, ROSA's illustrator was selected and would be announced once her final artwork was ready. On March 27, ROSA's singing voice database was confirmed to be developed for Synthesizer V. This was chosen due to the synthesizer's ease of use and ability to use VSQx files. On April 2, a recruitment post for users interested in creating songs and talk segments with ROSA was launched. On April 3, HirotoP mentioned that the company was interested in collaborating with Internet Co., Ltd. in distributing ROSA since they distributed CUL's VOCALOID3 packages in 2011.

ROSA's audition was launched on April 17. The illustration was confirmed to be released soon. Details and scheduling for the audition were noted on ZAN-SHIN Co., Ltd.'s official website for ROSA. A preview of ROSA's face was shown on April 20. On April 28, results of the first screening for the audition were revealed to be China Kitahara, Shirayuki, Juudenki, Minori Ninomiya, and Yuka Hiyamizu. ROSA's lineart for both Synthesizer V and CeVIO AI were revealed on May 1 and the colored sketches (rough illustrations) were released on May 4.

On May 8, Shirayuki was confirmed to be the chosen voice actress. Recording sessions for ROSA began on May 14. On May 15, Yoshida Iyo was confirmed to be the illustrator and the final artwork was shown. Information about ROSA's crowdfunding campaign in collaboration with SSS LLC. was revealed and was confirmed to run from June 1 to June 30 on GREEN FUNDING. The production project for ROSA's CeVIO AI talk voicebank would begin afterward and the project's page would be published as soon as it was ready. At that time, ZAN-SHIN was considering a system that would allow overseas fans to participate. The release order was confirmed to be the following: Japanese CeVIO AI Talk, Japanese Synthesizer V AI, English CeVIO AI Talk, and English Synthesizer V AI.

The start date for the crowdfund campaign was later changed to June 3 and procedures for overseas investors were being worked upon but was confirmed that this may be delayed. The goal was set to 7 million yen. The crowdfund for ROSA's CeVIO AI Talk voicebank began on June 3 with six tiers for Japanese supporters and three tiers for overseas supporters. A sample of Shirayuki voice-acting as ROSA was made available in the campaign to give supporters an idea of what ROSA may sound like prior to creating her voicebank. The CeVIO AI voicebank was fully funded on June 30, with only a few hours before the campaign's end, resulting in being 112% funded. Test recordings of the voicebank began on July 22, with ZAN-SHIN fully focusing on the production of the CeVIO AI Talk voicebank at this point in time.

On November 24, HirotoP created a demo song for Tsuina-chan's AI voice database. In a reply to the demo post, ROSA's official account noted that it may take some time before she would be able to talk to her and Koharu Rikka, then asked if they could sing together next year. When her CeVIO AI Talk Voice was announced to be releasing on December 17, ROSA's Twitter shared transparent versions of her CeVIO and Synthesizer V key artwork. On the Synthesizer V version's post, it was noted that ZAN-SHIN planned to have it created in 2022.

2022

On June 3, 2022, ROSA tweeted that she and WhiteCUL would be performing at the 柚子花主催LIVE -Planet Station- STAGE.3 online concert on June 19, however, WhiteCUL's character design would not be shown. On June 15, the lineup revealed that ROSA and WhiteCUL were expected to perform second, with the both of them singing. On the day of the concert, it was later clarified that the singing voices for ROSA and WhiteCUL were performed by their voice providers rather than using singing voice synthesizers. HirotoP hoped that the performances would give listeners an idea of what an AI singing voice database would sound like. It is currently unknown if WhiteCUL would be receiving a singing voicebank as, at the time, only ROSA was confirmed to be receiving a Synthesizer V AI voice database. On September 27, in response to a fan's question about the status of ROSA's Synthesizer V AI production, ZAN-SHIN confirmed that it had not started at the time but was still planned to be made.

2023

On May 31, 2023, Shirayuki tweeted that she had spent the past couple of days listening to and recording three songs, which started since 14:00 JST. She noted that "it reminded me of the nightmare of recording the talk software". However, it was not specified what the recordings were for.

On October 23, after an apology for the delays, ZAN-SHIN tweeted that the crowdfund for ROSA's Synthesizer V singing library would start in mid-November. The crowdfund was later pushed back to January or February of next year; AH-Software Co. Ltd. was also mentioned to be handling the production of her singing voice database.

On November 21, GUMI Radio announced (and HirotoP quoted) that Internet Co., Ltd. will handle the packaging and distribution of Synthesizer V ROSA in both physical-package and digital-download versions.

2024

On January 30, 2024 during the 13th official GUMI Radio broadcast, hosted by GUMI, Otomachi Una, Camui Gackpo, and kokone, a sample of ROSA's developmental stage voice database was shown.

On February 2, the crowdfund campaign for ROSA's Synthesizer V AI voice database production project was announced to start the next week. Start date and time for the crowdfund were later specified to be February 7 at 21:00 (9:00 PM) in Japan Standard Time, and the platform chosen for it was Ubgoe. For overseas supporters, special courses accepting PayPal payments were created (as Ubgoe only directly supports credit cards issued by Japanese banks); for courses involving physical goods (as well as the physical edition of the voice database), overseas customers were also asked to register with an international mail-forwarding proxy service like Tenso (because Ubgoe only accepts Japanese addresses, even for non-physical courses); a later tweet clarified that the overseas download course did not require entering any valid address, encouraging users to support it as is without using any forwarding company since they "do not acquire addresses". End date was specified to be February 28 at 23:59 (11:59 PM) JST.

When asked by a follower, ROSA's Twitter account was unable to know if she would receive a Lite version of her voice database.

The crowdfund ended with the support of 251 backers (including 22 overseas backers) pledging 5,171,950 yen, almost 74% of the original goal of 7 million yen.

Voice Database Information

Demonstrations

Demonstrations

AI Megpoid Official Demo	Twitter
Go Hard - Beta voice database	Niconico YouTube
さくら さくら (Sakura Sakura) - Beta voice database	Niconico YouTube
蒼い時間 (Aoi Toki) - Beta voice database	Niconico YouTube

Voice Databases

Product Information

Versions

  Lite Version/FLT Available?: TBA  Starter Pack Available?: TBA  Web Version Available?: No

Package details as noted:

ROSA is expected to be a feminine Japanese voice database based on the singing style of Shirayuki. As expected of an AI voice database, ROSA would be capable of singing with a more human-like and realistic voice.

Vocal traits as noted:

With Diffusion Probabilistic Models (DPM), ROSA would be able to sing more realistically through changes in volume and breath, further enriching singing details, making each singing unique, and providing creators with more options. Through this mode, she would be able to use AI Retakes to create multiple "takes" of a sung section to find an ideal vocal without the need for "laborious tuning". This feature is only available in the Synthesizer V Studio Pro Editor.

Phonetic notes as noted:

• Phoneme format: Romaji
• Through Cross-lingual Singing Synthesis, ROSA is expected to sing in not only Japanese but also in English, Mandarin Chinese, and Cantonese Chinese. This feature is only available in the Synthesizer V Studio Pro Editor.
  • A dedicated English voice database was proposed prior to the addition of the Cross-lingual Singing Synthesis feature. It is currently unknown if ROSA would forgo the dedicated separate English library and would either be trained with English songs to enhance English Cross-lingual Singing Synthesis or would not be trained with English data to instead rely entirely on the feature through Japanese samples alone.

Vocal Modes as noted (Synthesizer V Studio Pro Editor only):

• ROSA is planned to come with five variations:
  • Cute
  • Fun
  • Power
  • Soft
  • Solid

Voice Libraries

• ROSA Talk Voice (CeVIO AI), December 17, 2021
• ROSA AI (Synthesizer V Studio), In development, expected June 2024
• ROSA English Talk Voice (CeVIO AI), In consideration
• ROSA English AI (Synthesizer V Studio), In consideration

Additional Information

Marketing

INFORMATION PENDING

Reputation

ROSA's character status

Miscellaneous

• June 3, the day that ROSA's CeVIO AI crowdfund was launched, was acknowledged as "ROSA Day".
• ROSA's birthday, December 17, is also the day she released for CeVIO AI in 2021.

External links

• ZAN-SHIN Co., Ltd. Official Twitter
• ZAN-SHIN Co., Ltd. Official ROSA Website
• Official ROSA Twitter