The present research work seeks to develop a security system for the control of access to a dwelling by voice recognition of the speaker dependent on the text, where the main objective is to identify each member of the household, who will have access. The system consists of 2 phases, the first is the extraction of sign al characteris- tics, for this an adaptation of the MFCC was used and the second is the automatic recognition of the speaker that is based on the DTW technique, in addition to a logic for decision making on the identification or not, or false identification of a user. In relation to the configuration of the experiments, to verify that the system is reliable, 2 types of tests were carried out, one among the users of the system trying to access one with the user of the other and the other where people outside those stored in the database (intruders) tried to access. As a fin...