is basically to identify a signal based on a short sample for it which usually has its intrinsic features and thus these intrinsic features
Each group will collect N different songs with something common or similar among them (e.g. same singer voice, or same lyrics, or similar tunes), where N is the number of group members.
Extract each song into two separate files (music & vocals) and upload it to the common repository. You are not requested to do the separation by code. This is an advanced topics of components separation that can be found in higher level DSP courses (Blind source separation, ICA topics for those who might be interested to
#also There should be a sheet that contains the songs names to avoid repetition. When you get your songs, put their names along with their similarity feature in the sheet. Anyone who will put the same song later will be penalized.