Makes instrumentals using MVSEP, usually with htdemucs_ft (High quality Demucs4) model. My channel is a "demo" of music/voice separation and MVSEP itself, demonstrating how the best AI models available can make very good instrumentals with some distortion. I am currently also developing a website where the instrumentals are going to be displayed, you can also play the vocals track at the same time or choose between the two (like in Audacity where you can mute individual tracks). The website is also going to have a live lyrics section for the particular song.
Here you can submit song requests: forms.gle/gUPQ4Ng1fH9U6Bux7