The purpose of this research is to support office workers to participate in multiple teleconferences simultaneously. In order to achieve this goal, we have investigated how people understand multiple voices that differ in conditions of overlapping rates. We have evaluated comprehension of the context and the keywords in multiple voices, which is necessary for the users to participate in multiple teleconferences. In addition, we have described the psychological load of the users by using NASA-TLX as the workload index and the physiological load by examining the brain waves of the users. From the experiment, we can show three factors. First, we found more than half of the examinees understand the context when the voices are overlapped completely. Second, little of no difference is observed in the level of comprehension of keywords, between when the voices are half overlapped and overlapped completely. Third, it can also be suggested that examinees are more uncertain of their answers when the voices are overlapped completely compared to when they are only half overlapped. As for the load of the users, our results suggested that imperfect overlap amplifies the psychological load. Based on these results, we will discuss the necessity of selecting appropriate overlap rates and design the environment of multiple teleconferences.