This paper presents a novel speak-to-VR virtual reality peripheral network (VRPN) server based on speech processing. The server uses a microphone array as a speech source and streams the results of the process through a Wi-Fi network. The proposed VRPN server provides a handy, portable and wireless human machine interface that can facilitate interaction in a variety interfaces and application domains including HMD- and CAVE-based virtual reality systems, flight and driving simulators and many others. The VRPN server is based on a speech processing software development kits and VRPN library in C++. Speak-to-VR VRPN works well even in the presence of background noise or the voices of other users in the vicinity. The speech processing algorithm is not sensitive to the user’s accent because it is trained while it is operating. Speech recognition parameters are trained by hidden Markov model in real time. The advantages and disadvantages of the speak-to-VR server are studied under different configurations. Then, the efficiency and the precision of the speak-to-VR server for a real application are validated via a formal user study with ten participants. Two experimental test setups are implemented on a CAVE system by using either Kinect Xbox or array microphone as input device. Each participant is asked to navigate in a virtual environment and manipulate an object. The experimental data analysis shows promising results and motivates additional research opportunities.