In this study, we aimed speech recognition-based remote control of home devices. The system was designed at two stages: speech recognition module and transmitter module that sends commands including fewer than two main headings. In the first stage speech recognition algorithm is implemented in MATLAB. Mel Frequency Cepstrum Coefficients (MFCCs) are used as a feature extraction method. Dynamic Timing Warping (DTW) is used as a feature matching method. After the recognition of speech commands entered, Arduino pin situation is changed via USB. The voltage change in pin is detected by the same pin connected to the transmitter Zigbee module. When the change is detected, Zigbee transmitter sends data to connecting on the desired device to be controlled the receiver Zigbee module via wireless network. When sending data detected by the receiver Zigbee module, 5V relay is triggered connecting on the device pins of the module. Thus, it was possible to successfully remotely control any device with speech commands using wireless network.