Please refer to the French terms of use for legal information.

The following terms of use govern the use of ALYS and LEORA for DiffSinger, voice banks developed for DiffSinger by Utau France in collaboration with the ALYS copyright holders.

By downloading ALYS and/or LEORA for DiffSinger, you agree to abide by the terms of use described below.

I. CHARACTERS

We distinguish between the ALYS and/or LEORA voice bank (software) for DiffSinger characters.

The ALYS and LEORA characters, created by former VoxWave members, may not be used for commercial purposes without the authorization of the relevant copyright holders. ALYS is a registered trademark of Cyrielle Collignon, LEORA is a registered trademark of Siavosh Kayvantash.

II. SOFTWARE

ALYS’s voice for DiffSinger is based on that of the singer Poucet. It cannot be attributed to another person.
LEORA’s voice for DiffSinger is based on that of the singer TBK. It cannot be attributed to another person.

Commercial use of ALYS and/or LEORA for DiffSinger is strictly forbidden without Monte Cristo Collective authorization.

Redistribution of ALYS and/or LEORA for DiffSinger on platforms not authorized by the copyright holders is prohibited.

Reverse engineering of ALYS and/or LEORA for DiffSinger is strictly prohibited.

The use of ALYS and/or LEORA for DiffSinger in any context other than the creation of covers on OpenUtau (or other interface for DiffSinger) is strictly forbidden. This includes in particular the use of AI voice modifiers (Diff-SVC, RVC etc.) in training.

Porting ALYS and/or LEORA for DiffSinger to other software is forbidden without the prior authorization of the copyright holders.

Any misuse of ALYS and/or LEORA for DiffSinger may result in the withdrawal of its distribution.

The copyright holders reserve the right to modify the conditions of use and distribution of ALYS and/or LEORA for DiffSinger at any time.

The copyright holders reserve the right to demand the withdrawal of any abusive use of ALYS and/or LEORA for DiffSinger.

The creation of content with ALYS and LEORA for Diffsinger must comply with current French legislation, proh

III. CREDITS

The ALYS port for DiffSinger was made by Utau France from the original ALYS samples.

The porting of LEORA for DiffSinger was done by Utau France using the original LEORA samples and new recordings by TBK.

Credit ALYS  for DiffSinger

S’pose (imsupposed2) – Audio standardization, data processing, Japanese labels and DiffSinger training

Gyromancy (GyromancyX) – Quality assurance, tester, logistics

Hibya (Tweek_Twix) – Website, quality assurance, tester, logistics

Mim (mimsynth) – French labels, quality assurance, tester, logistics

ALYS team – Quality assurance, coordination, rights management

Poucet – ALYS doubler

Crescentia Studio (CrescentiaUTAU) – Piano roll/Official art
Saphirya – ALYS web page illustration

Credit LEORA for DiffSinger

S’pose (imsupposed2) – Audio standardization, data processing, English and Japanese labels and DiffSinger training

Gyromancy (GyromancyX) – French labels, quality assurance, tester, logistics

Hibya (Tweek_Twix) – Website, quality assurance, tester, logistics

Mim (mimsynth) – French labels, quality assurance, tester, logistics

ALYS team – Quality assurance, coordination, rights management

TBK – LEORA doubler

Crescentia Studio (CrescentiaUTAU) – Piano roll/Official art

Elairin – Illustration of LEORA web page

IV. Multilingual support (DiffSinger training)

Due to the Millefeuille model using data from many voice providers the following standardized Terms Of Service apply to all characters:

All models and model assets created and trained by UFR will be referred to as “Resulting models”

– Resulting models must be free upon distribution, selling or use for trades is prohibited.
– No AI generated art or video may be used for voice bank assets, covers, PVs, lyric videos, album covers, or other media depicting the resulting model.
– No AI generated music may be created with the resulting model
– Commercial use of renders or model assets for advertisements or products not including the sell of doujin music is prohibited.
– The use of renders or model assets for creating new products, corpuses, or AI training is prohibited.
– Resulting models may not be used for illegal activity.
– Resulting models may not be used in depictions of non-consensual relations.
– Resulting models may not be used for hate speech.
– Individual character TOS may add more restrictions but cannot override these basic protections outlined in the UFR Multimodel TOS.