To estimate TDOA by Kmeans , and then use the LCMV-GSC to do speech enhancement or BSS. This idea raises when I was learning Kmeans in ML.Then I realized Kmeans could be used to classify TDOAs in non-overlapped frequency. There maybe some bugs in this code,and many improvements on both code and algrithem could be done. If u think it's useful,please refer me.
srmchcy/TDOA-estimated-by-Kmeans
To estimate TDOA by Kmeans , and then use the LCMV-GSC to do speech enhancement or BSS.
MATLAB