This folder contains the audio samples of the SepFormer without overlap. The files are named the following way:
- mix is the input mixture
- s1 is the clean version of the first source
- s2 is the clean version of the second source
- e1 and e2 are the outputs of the model (meaning the estimations of s1 and s2)

Since permutation is random, sometimes e1 is an estimation of s1 and sometimes e1 will be an estimation of e2.
