Speaker recognition becomes the most popular task these days. Many methods and features use in speaker recognition. MFCC(Mel-frequency cepstral coefficients) is the most powerful feature. SVM(Support-vector machine) ware used with MFCC feature. First I tried for five speakers and after that, I did for 50 speakers. This method is working pretty well for 50 speakers.
to know more about MFCC:- https://ieeexplore.ieee.org/document/5709752
for more information about SVM:- https://en.wikipedia.org/wiki/Support-vector_machine