Technology is growing so fast. One of that technology is voice recognition. It has been implemented in the electronic equipments, buildings security or intelligence houses. This final project discusses how human vowel voices can be recognized and implemented to activate lamps. The system uses some available software to manipulate data such as Matlab, Visual Basic and Assembly language. Sample voices had been taken 50 times from man and woman through microphone. After doing some calculations of taken voice such as filtering, doing Fast Fourier Transform, the result is processed through artificial neural network for identifying the meaning of that voice. The output is transferred to the other hardware using DT-51 Low Cost System and relay board to activate the certain lamp. Testing is also had been done. Even though the system can not identify other voices that had not been taken before, it can recognize the voices taken quite well. Its accuracy achieves about 98.5 %. Threshold that is used for testing is 0.945366244 which is determined by analyzing the results for each of neural network configuration.