IA (VoiSona)

This article is about the VoiSona software known as a voicebank. If you are looking for the VoiSona character then click here.

History

On January 27, 2024, 1st PLACE Co., Ltd. and VoiSona's Twitter accounts jointly announced that IA AI SONG, IA AI SONG English, and ONE AI SONG were now available on VoiSona. It was also announced that owners of their respective CeVIO AI Song Voices will be able to download and use their VoiSona versions free of charge, only requiring that they input their CeVIO AI license serial numbers after logging into their VoiSona accounts; no credit card information is required from them to register.^[1]^[2]^[3]

On August 1, versions 2.0.1 of IA AI SONG, IA AI SONG ENGLISH, and ONE AI SONG were released for VoiSona, using a new type of vocoder that first debuted on VoiSona; while IA AI SONG did also receive a corresponding 2.0.0 update for its CeVIO AI version, IA AI SONG ENGLISH and ONE AI SONG did not.^[4]^[5]

VoiSona

VoiSona IA AI SONG

Product Information

A port of CeVIO AI's IA AI SONG voice library (version 1.0.1) for VoiSona. A feminine vocal capable of Japanese singing synthesis.

Demonstrations

Samples


花緑青 (Hanarokushou)	YouTube
時を刻む唄 (Cover)	YouTube
クラウドライダー (Cloud Rider) (Self-Cover)	YouTube

Note: As the VoiSona version of IA produces the same singing results as the CeVIO AI version, the demonstrations used to showcase the product are the same.

System Requirements

Supported Operating Environment
- Windows 11 Japanese / English edition (64bit)
- Windows 10 Japanese / English edition (64bit)
- macOS 10.12 or later
CPU: Intel / AMD dual core processor or higher (4 cores or higher recommended)
Memory: 4GB or more (8GB or more recommended)
HDD: 1GB or more
Display: 1280 x 720 or higher full color
DAW: DAW software compatible with VSTi that generates 48kHz waveforms
Other: Internet connection is required for license activation and update.

VoiSona IA AI SONG ENGLISH

Product Information

A port of CeVIO AI's IA AI SONG ENGLISH voice library (version 1.0.2) for VoiSona. A feminine vocal capable of Japanese singing synthesis.

Demonstrations

Samples


INTERGALACTIA (MV)	YouTube
INTERGALACTIA (Teaser)	YouTube
REZONA , Into the night[feat.HIPPI] (references REZONA and Into the night)	YouTube

Note: As the VoiSona version of IA produces the same singing results as the CeVIO AI version, some of the demonstrations used to showcase the product are the same.

System Requirements

Supported Operating Environment
- Windows 11 Japanese / English edition (64bit)
- Windows 10 Japanese / English edition (64bit)
- macOS 10.12 or later
CPU: Intel / AMD dual core processor or higher (4 cores or higher recommended)
Memory: 4GB or more (8GB or more recommended)
HDD: 1GB or more
Display: 1280 x 720 or higher full color
DAW: DAW software compatible with VSTi that generates 48kHz waveforms
Other: Internet connection is required for license activation and update.

Voicebank Libraries

Product Information
Engine as noted:
VoiSona Song	Optimum
Language: Japanese Optimum Range: B3-G5 Optimum Tempo: 70-130BPM
Package details as noted:
Released 'January 27, 2024' by 1st PLACE Co., Ltd., this voicebank's stable release is VoiSona, ported from version 1.0.1 of her CeVIO AI Song Voice. VoiSona IA AI SONG produces the same voice output as CeVIO AI IA AI SONG.	Trial Version/Starter Pack
Trial Version: No
Phonetic notes as noted:
This voicebank was recorded for the Japanese language. Her voice sounds quite different from her VOCALOID variants when utilized in its raw state; she has a different tone and seems younger. The AI treatment could be to blame for this.
Software issues as noted:
[All versions] In contrast to her VOCALOID voicebanks, IA lost two notes at the top of her range (her V3 had a range of B2-A4, which equates to B3-A5, currently her upper recommended note is only G5) and a significant portion of her tempo range (63-228BPM for the original V3 bank to only 70-130BPM). When comparing AI voicebanks to their more "standard" equivalents, this is something that is discovered to be rather frequent. When left unedited, this causes her to stumble over quick tempos and results in a "slurred" performance. This is fixed, though, by manually shifting the vowels' timing such that they start at the beginning of the note. [1.X] IA occasionally pronounces some phonemes incorrectly in notes when they are left unedited and use standard phonemes, particularly when the phoneme ends in "e". For instance, IA can sing "te" as "ta" or "de" as "da". This is fixed by adding a brief [h] note at the end (1/32q).
Voicebank sample
Solfege: 1.0.1: 2.0.1:
view • edit

Product Information
Engine as noted:
VoiSona Song	Optimum
Language: English Optimum Range: F3-F5 Optimum Tempo: 70-170BPM
Package details as noted:
Released 'January 27, 2024' by 1st PLACE Co., Ltd., this voicebank's stable release is VoiSona, ported from version 1.0.2 of her CeVIO AI Song Voice. VoiSona IA AI SONG ENGLISH produces the same voice output as CeVIO AI IA AI SONG ENGLISH.	Trial Version/Starter Pack
Trial Version: No
Phonetic notes as noted:
This voicebank was recorded for the English language.
Voicebank sample
Solfege: File:IA AI SONG ENGLISH VoiSona 1.0.2 ported from CeVIO AI.ogg
view • edit

References

Navigation

[1] ttps://x.com/1st_PLACE0302/status/1751228255729013137

[2] ttps://x.com/VoiSonaOfficial/status/1751228562961506807

[3] ttps://voisona.com/news_detail/55/

[4] ttps://cevio.jp/guide/cevio_ai/home/ia_song_update

[5] ttps://x.com/VoiSonaOfficial/status/1818844114772754589

[1]

[2]

[3]

[4]

[5]