This paper presents the results obtained in the comparison of several Artificial Neural Networks for phonetic speech recognition and speaker recognition tasks. Both processes are very important in the human-computer interface through telephone systems for automatic communication between a customer and an automatic receiver system such as an automatic dialing system. Speech samples stored from a standard digital telephone exchange system are being used to test several Artificial Neural Network models already trained using different features from the speech signal. Comparative results of all these experiments are presented, together with several conclusions obtained from these results.

Keywords: Neural Networks, Speech and Speaker Recognition