As jcwren says above, DFT is probably the way to go. Beware, however, that your clips have the same length. If you have the same sound that in one sample is a 1/4 of a second longer than the other, and your sample rate is 1/2 a second, then you will be taking completely different sample points and your algorithm will identify the sounds as different. That may be construed as a feature, however.