README.md

metadata

title: Vocal Separation SOTA
emoji: 🎤
colorFrom: red
colorTo: gray
sdk: gradio
sdk_version: 6.3.0
app_file: app.py
pinned: true
license: mit

Vocal Separation SOTA

This is a demo for SOTA vocal separation models. Upload an audio file and the model will separate the vocals from the background music.

Based on the result of MDX23, the current SOTA model is BS-RoFormer.

For comparison, you can also try the Mel-RoFormer model (a variant of BS-RoFormer) and the popular HTDemucs FT model.

Models

• BS-RoFormer
• Mel-RoFormer
• HTDemucs FT

The models are trained by the UVR project.

Based on JacobLinCool/vocal-separation.