This paper presents a voice user interface consisting of several modules for a mobile service robot, which is used to guide people and provide information on a university campus. The recognition and processing system is based on cloud services to convert from speech to text and vice versa and a dialogue system to allow for natural interaction. An approach to combine these modules with a data management system for meal plan, public transit, and location information is presented. We evaluate the system in different environments, each with their individual reverberation times, proving the functionality under conditions typical for the intended use case. In a user study with 13 participants we show the usability of the system, by letting the participants freely interact with the robot. In 86 % of all cases the desired output can be achieved at least once per user and request. A questionnare shows that most users agree with a good usability of the system.