Vocal Synthesizer Wiki

KAFU (()()) is a Japanese CeVIO AI voicebank by KAF made in collaboration with KAMITSUBAKI STUDIO as part of th ARTIFICIAL SINGING SYNTHESIZER series. She is capable of singing.[1][2] KAFU was released in July 2021.

Her voice provider is the Virtual Singer and YouTuber, KAF.


KAFU is regarded as a "ARTIFICIAL SINGING SYNTHESIZER" (音楽的同位体) of KAF.[3] This is likely done as a way of separating KAFU from KAF.[4]


KAFU's name is a combination of "KAF", her voice provider, and "You", stylized as "U".


KAFU's design is based off of her voice provider, KAF. She wears a heavy white jacket with blue and black detailing and a hood similar to that of KAF’s own attire and an over-the-knee length black dress. She wears white tights, yellow socks and blue heeled short boots. Her hair is snow white, and bobbed with low pigtails on either side reminiscent of KAF's hairstyle and is adorned with large diamond shaped hair clips. KAFU’s eyes are dark blue and purple.


  • KAF - Voice Provider and basis for KAFU's character.


On October 10, 2020, it was announced that KAF would be collaborating with CeVIO AI. Her voicebank was announced at that time to be distributed in the winter of 2020.[1]

A demo for KAFU was uploaded on October 13, 2020[5] and a YouTube channel was created for her.[6]

On October 15, 2020 a questionnaire was announced that would be used for determining the type of voice KAFU would have.[7] An accompanying video showcasing three potential vocals was also uploaded including: TYPE // A, TYPE // B and TYPE // C. Users were able to voice their preference for whichever vocal they liked the most.

On October 18, 2020 the results of the questionnaire were tallied with TYPE // A receiving a rating of 65% of the total votes, TYPE // B receiving a 25.3% of the total votes and TYPE // C receiving a 9.7% of the total votes. Here, it was announced that the voice provider KAF would be making the final decision of which vocal would be moving forward with development with reference to the survey results.[8]

On October 19,2020 KAF herself followed up with a tweet regarding her own feelings on KAFU. KAF stated that while she was overall thrilled of the vocal as a fellow VOCALOID fan herself, KAF believed that the TYPE // A voice was too similar to her own vocals and was concerned that KAFU would "replace" her, thus making the final decision that TYPE // B would be the vocal to go into development.[9][10]

After the the conclusion of the questionnaire results, KAFU's twitter announced on October 27, 2020 that the development period would be extended with a delayed release date in hopes of satisfying both KAF herself and fans. KAFU was slated to be released within the first half of 2021[11]

The first full demo for KAFU was uploaded on November 14, 2020 titled "雛鳥" a cover of one of KAF's own original songs.[12]

On January 21, 2021, the official website for KAFU was created and revealed her full illustration as well as information regarding pricing, purchase options and locations. It was revealed KAFU would be releasing as starter pack with both the Song Voice and CeVIO AI editor as well as an original compilation album or the Song Voice being sold separetely.[13]

On March 14, 2021, KAFU was released for pre-order.[14] A special retweet campaign was announced to commemorate KAFU's release in which 10 users would be selected to receive KAFU's CeVIO AI Starter Pack. [15]

On June 24, 2021, KAFU's terms and guidelines for her character and voicebank usage were posted to her website as well as the official KAMITSUBAKI STUDIO website. [16][17]

She was finally released on July 7, 2021, her download version having been available as well on Vector Shop.





In October 2020 a questionnaire was released to determine what voice type would be used for KAFU.[7] Three types of samples were releaed with TYPE // A receiving the majority of votes with 65%.[8] However, the next day KAF tweeted that TYPE // B would be the voice going into production instead due to feeling that TYPE // A was too similar and fears that KAFU would end up replacing her.[9] The response to the results were mostly positive and understanding, but there were some who were upset that the final results did not reflect the voicebank chosen.




External links[]