The growing maturity of autonomous driving technology has enabled designers to study how to give car owners better driving experience. However, the study of the auditory interaction between the pedestrian and the autonomous car is still not known sufficiently. The goal of this research is to study the user experience, real emotional feedback of the pedestrian for alarm sounds from the autonomous car. The whole research included one online survey and two experiments. There were three samples of sound used in this study: engine sounds, bells, and soft alarms. Experiments were conducted with factors such as sound preferences, the risk of sound, and different road environments analyzed the differences in pedestrian intelligence processing. The results of the user preference showed participants preferred soft sounds and noticeable sounds. However, pedestrians can feel sounds convey degrees of risk, but sound conveys the level of risk that is not affected by the environment.