One of the most term in the biometric technologies is the voice recognition. Speech recognition is a complicate problem .But in today's society when technology is consistently striving for voice driven implementation, speech recognition can be a very useful system .This paper aims to build a system of voice recognition using two systems one of them is based on Fourier descriptors of speech and the other is Artificial neural network which is using back propagation algorithm in neural networks, The first speech recognition system is using Fourier descriptor of speech to get four formants of any speech signal. The second intelligent system is built neural network which is trained and tested using data sits of formants of speech signals. After we applied two systems for speech, the first method is not accurate because it depends on the human element in comparison and identification. This method makes it possible to have the error ratio greater than the second method, which is smart and accurate and depends on the intelligent system stable and continuous and has no life As well as not influenced by the external influences that might control the decision in the first method.