This repositery gathers the elements used for the audio processing hackathon organized by Automatants in 2023. The goal was to create a generative model able to split an audio consisting of two people talking at the same time into two audios of the corresponding voices. I teamed up with two other students and we managed to claim the 3rd place of the leaderboard.