Carnatic singing voice separation trained with in-domain data with leakage:
audio examples

From the paper "Carnatic Singing Voice Separation Using Cold Diffusion on Training Data with Bleeding"
G. Plaja-Roglans, M. Miron, A. Shankar and X. Serra
Paper accepted for presentation at ISMIR 2023, Milan, Italy

Link to paper

GitHub code

Example 1

Original mixture

Our model

Spleeter

Example 2

Original mixture

Our model

Spleeter

Example 3

Original mixture

Our model

Spleeter

Example 4

Original mixture

Our model

Spleeter

Example 5

Original mixture

Our model

Spleeter

Example 6

Original mixture

Our model

Spleeter