This paper presents an experimental study of the implementation of speaker verification on a mobile embedded system. Specifically, our speaker verification system is built using the mel-warped cepstral feature extraction and GMM-based pattern matching methods. We outline in a step-by-step method the procedure for porting such a system from a desktop PC to a PDA. Experiments show that our system runs 37 times faster than the baseline system so that speaker verification becomes feasible on a PDA.