Let me explain, I am calculating speakers from voice, for example i have 10 test clips. Each clip is named with a person, now my Machine Learning algorithm identifies the people of that clip.

Like, my actual clips = {A, B, C, A, C, B, B, B, A, B} and Identified Class = {A, C, C, A, C, B, A, B, A, C}, that means my algorithm identifies 7 clips properly.

Now, I want to calculate the Precision, Recall, Accuracy etc. from this data.

However, I am confused about what is the True Positive (TP) class here and what are True Negative (TN).