Serj Tankian (2001, Toxicity Era)

Description

New Toxicity era model! The current AI source separation models let you create better datasets compared to even last year, so I decided to update this specific model. Compared to the previous version, this model is better at angry/distorted vocals, while also doing well on clean and quiet singing (it struggles a bit with high-pitched singing, keep that in mind). Hope you enjoy using it!