TIGER for DiffSinger v106

Changes:

• Korean XLS (Use Default phonemizer and enter in hangul characters!)
• Trained with "Reflow" instead of "ddsp", overall better performance and higher quality
• Better pitch models

Download and unzip the "pack" file, and you will find a guide on how to install!

TIGER for DiffSinger v103

🐯 TIGER for DiffSinger version v103!!!

To install, decompress "Pack" folder and follow directions listed in the readme file!

Changelog:

• Removed "Cetera" voicemode
• Added "Glam" voicemode
• Renamed "Falsetto" voicemode to "Royal"
• Added support for French!
• Reconfigured pitch models, tiger now has 3 instead of 1 (Fresh/Vinyl/Royal, Disco/Mystic, Electric/Glam)
• Added dsdict's for french, chinese mandarin and portuguese.
• Currently, only french is 100% supported, but more to come.

TIGER for DiffSinger v102

TIGER DS v102 - Released Jan 22 2024
🐯

Changes:

• Removed Chinese voice mode for the time being
• Retrained Acoustic model on standard hyper parameters.
• Better XLS Support for Japanese and Spanish, native data in the multispeaker model now.
• Supports "DIFFS ES" Fully
• Updated, super good auto-pitch tuning model.
• Pack includes custom phonemizer plugins

TIGER for DiffSinger v101

This is a voicebank optimized for use in OpenUTAU!

To install, download TIGER_DS_v101.zip and drag-and-drop it into OpenUTAU.

TIGER DS v101 🐯

• Higher quality acoustic model using custom hyperparameters. The acoustic model is about 100mb larger than the standard acoustic model hyperparameters.
• Not-completely-implemented Mandarin Chinese support, with the added phonemes qz jz xz rz ez ts dz ir iz yu ye. Currently, the only way to utilize these phonemes is using the DIFFS phonemizer and entering them manually, but will be properly implemented in a future version.
• More accurate Japanese phonetics using the borrowed phonemes from Chinese: ts dz.
• Fine-Tuned pitch model provides much more stability to generated pitch than previous versions.