Can the accuracy of NN classifier be 100% for training and validation?

I am working on NN toolbox to classify my given speech emotion samples(total 96 speech samples, 6 samples for each emotion) into 4 classes of emotions. I have used the MFFC for feature extraction and created a matrix of 8450 X 96 to be used as input matrix and a target matrix of 4 X 96. when start my NN train it gives me the following error % I am not very sure whether getting this 100% accuracy in training and validation is really possible. please advice me if i am going wrong anywhere.