Mai is a Japanese vocal developed and distributed by Dreamtonics Co., Ltd., and released as an AI voice database for Synthesizer V Studio in November 2022.
Her voice provider is Honoka Kitta, a Japanese singer-songwriter.[1]
Concept[]
Like previous Dreamtonics vocals before her, Mai was intended to be treated as an official voice database or an indispensable "reference-level" type of singing voice, representing Dreamtonics's brand quality.
Mai was designed to have "crisp pop singing vocals" suitable for J-Pop, Dance, and Idol styles of music. The characteristics of her voice is retained throughout her range, allowing her to keep "the vibe of the song" undisturbed whenever she changes keys.
Etymology[]
No official information was released regarding why the name "Mai" was chosen or what the name may mean.
Appearance[]
Like previous Dreamtonics vocals, rather than having an official human-like avatar, Mai is represented by an abstract symbol, the letters M and A, which are drawn from the first two letters of her name. The letters are light yellow in color with hints of magenta. The top of the M's arches are flat rather than pointed. The "A" is shorter than the "M" and is more of a box-shape rather than pointed. Characterization is left to the song producers themselves rather than the product, so Mai is open to any form of visual interpretation, allowing the producer to decide what they would like her to look like.
Voice Libraries[]
- Mai AI (Synthesizer V Studio), November 10, 2022
Additional Information[]
Marketing[]
Like previous Dreamtonics vocals, Mai was treated as an official or standard vocal for the engine, where she could potentially be found in demonstrations of Synthesizer V Studio and its AI capabilities. This also meant that she would be among the first vocals to be equipped with the latest engine updates. Mai was the focal voice of marketing Synthesizer V Studio 1.8.0, which introduced Diffusion Probabilistic Models (DPM), MIDI keyboard and Metronome support, and Note-level cross-lingual synthesis. She was featured in the introduction video for the 1.8.0 update, specifically the sections that featured MIDI keyboard and Metronome support and the Note-level cross-lingual synthesis function (with the YouTube version featuring her in English and the bilibili version featuring her in Mandarin Chinese).[2][3]
Similarly, Dreamtonics began uploading tutorial videos that were suited for new users of Synthesizer V Studio since October 17, 2022. Mai's voice was featured in the YouTube versions of the series' third[4], fifth,[5] and sixth[6] videos.
On November 10, 2022, Mai was released as a free vocal for all Synthesizer V Studio Pro users, a first for the Dreamtonics full voice database line-up.[7] Through an interview with DTM Station, Kanru Hua explained that Synthesizer V Studio Basic and Lite versions of certain voice databases were released so users could experience it for free with limited functions. On the other hand, upon purchasing Synthesizer V Studio Pro, there weren't options like this for full voice databases, leading to the free distribution of Mai so users could experience the fun and wonders of AI singing voice synthesis.[8] After her release, certain Japanese music retail stores may have a trial of Synthesizer V Studio Pro and Mai available, where customers can use the software that was set up with a monitor and keyboard.[9]
On December 9, 2022, Dreamtonics announced that Mai, Saki, Qing Su, Ryo, Kevin, An Xiao, Feng Yi, Yuma, Natalie, Cong Zheng, and Xuan Yu officially have their own artist association profiles on NetEase Music. Creators can use the profiles to add to their published songs, providing a more unified retrieval method and higher exposure rate for the works. Dreamtonics would provide assistance services to creators at no charge and would not obtain any income belonging to the artist when associating their works with the singer profiles. Based on NetEase Music's policies, creators needed to provide relevant proof of purchase to NetEase Music's customer service in order to associate a virtual singer with their works, thus, screenshots of receipts would need to be prepared in advance. NetEase Music's customer service would then confirm the request and forward it to Dreamtonics, who would approve the requests within five business days upon receiving them.[10]
Initially unrevealed upon debut and release, on March 15, 2023, Mai was confirmed to be voiced by Honoka Kitta, and was among the few Dreamtonics vocalists with a revealed voice provider. Along with an interview, Honoka Kitta also released her own cover of Mai's original demo song "Merry-Go-Round" in celebration of the official reveal.[11]
Reputation[]
- Mai is currently the AI voice database with the least number of Vocal Modes available, with the total being two.
- The first song featuring Mai to reach one million views was a cover of "WĆ¹ LĒ" by Che Ling Miao, Mala Tu Tou-P, and Ze You io, which was achieved on January 12, 2023.[14]
- First full AI voice database to be released for free.
References[]
- ā https://twitter.com/dreamtonics_jp/status/1635899276810416129
- ā https://www.youtube.com/watch?v=KxwLaLn4zbY
- ā https://www.bilibili.com/video/BV1s84y1v7UJ/
- ā https://www.youtube.com/watch?v=oFkKvLSAQxo
- ā https://www.youtube.com/watch?v=y64Fryg9OxU
- ā https://www.youtube.com/watch?v=TNOLxr62Np0
- ā https://twitter.com/dreamtonics_en/status/1590600959725830146
- ā https://www.dtmstation.com/archives/58816.html
- ā https://twitter.com/shima_tendo/status/1606571112032522241
- ā https://weibo.com/ttarticle/p/show?id=2309404844858597835082
- ā https://www.youtube.com/watch?v=_W8x3bUIj5o
- ā https://www.dtmstation.com/archives/58686.html
- ā https://twitter.com/dreamtonics_en/status/1590600963978833923
- ā https://www.bilibili.com/video/BV1we4y1W7if/