Printto Magicbeat V5 Indigo  contains seven vocal modes—Default, Natural, Solid, Dark, Soft, Sweet, and Light. All samples were recorded from scratch with improved equipment.
Indigo offers a more professional and luxurious vocal.

Indigo Samples: JA EN TH

Printto Magicbeat V5 Scarlet  contains six vocal modes—Solid, New Casual, Soft, Adult, RE, and Alter. The samples were recorded with a Samson Go Mic.
Scarlet offers a more youthful and bright vocal.

Scarlet Samples: JA EN TH

Languages	English, Thai, Japanese, French, Russian, Korean, Spanish, Chinese, Indonesian, and Filipino
Vocal Modes	Indigo Default, Natural (English Ex), Solid, Dark (Japanese Ex), Soft, Sweet, and Light	Scarlet Solid, New Casual, Soft, Adult, RE, and Alter
System	Language Prefixes
Data Used	Alex Floarea Bento Fuwari Chinoame Chiyu CSD (Relabeled by heta-tan) FranceFrank Kaiz Keiro LIFE Llane Crow Kone Hiro Lucy Roos M4Singer (Relabeled by nobodyP)	Namine Criss Ofuton P Opencpop (Relabeled by Komisteng) Millefeuille PJS (Relabeled by Utautautau) Project AI❤dol Lucy Roos Tanya Magicbeat TGM Corpus Toshio Sasagawa Zephyr
Vocoder	pc_tgm_hifigan

Printto Magicbeat V5 Pure (also known as the Commercial Pack) was trained using only Printto Magicbeat vocal data, with no additional support from parallel training.

This voicebank offers the purest Printto Magicbeat vocal to date and supports Thai, Japanese, and English. It includes both Indigo and Scarlet, based on the V5 voicebank.

Commercial use of this voicebank/model is permitted in accordance with the Terms of Use.

Languages	Thai, Japanese, and English
Vocal Modes	Indigo Default, Natural (English Ex), Solid, Dark (Japanese Ex), Soft, Sweet, and Light	Scarlet Solid, New Casual, Soft, Adult, RE, and Alter
System	Language Prefixes
Data Used	No additional data used
Vocoder	pc_tgm_hifigan

Printto Magicbeat Millefeuille Edition is a DiffSinger voicebank trained by UFR. The model is based on Printto TH Dataset (v24.08.05). This voicebank supports 12 languages following Millefeuille's XLS standard.

Thanks to UFR’s training, this model features the most refined pitch model among all Printto Magicbeat voicebanks.

Languages	Chinese, English, French, German, Italian, Japanese, Korean, Polish, Portuguese, Russian, Spanish, and Thai
Vocal Modes	Default and Casual
System	Multi-dictionary
Data Used	Millefeuille XLS Credits

Printto Magicbeat Japanese ENUNU voice library for OpenUTAU.

Printto Magicbeat RE is a series of voicebanks reusing the samples originally recorded for his DeepVocal voicebank back in 2019. The samples were recorded with a high-quality setup.

The RE series is released in multiple formats, UTAU and DiffSinger. The Printto Magicbeat RE UTAU version is a standalone voicebank, while the Printto Magicbeat RE DiffSinger version is included as an expression in his main DiffSinger voicebank.

System	Japanese CVVC
Pitches	Multi-pitch (D3, F3, C4, and F4)
Vocal Modes	Default and Remove Falsetto

Alter Project is a project aimed at presenting another side of Printto Magicbeat. It was first demoed in DeepVocal in 2019 and has now been revived in 2025. The project explores different vocal styles of Printto Magicbeat that are not suited to the main character.

Please check the V5 voicebank.

This project was already merged to the main DiffSinger voicebanks.