With the advancement of artificial intelligence technology, it has become increasingly popular for intelligent agents to perform complex tasks such as fully autonomous driving. However, it remains unclear and needs further investigation whether intelligent agents’ voice modes present more assistance or risk to users. By using a within-subjects design, this study examined the effects of voice gender (male, female) and pitch (high, low) of in-vehicle intelligent agents on drivers’ likeability and trust in the system. The experiment was conducted remotely via an online platform, where participants viewed four recorded videos as simulated autonomous driving scenarios and completed a questionnaire after watching each video. There were 25 participants in total who participated in the study. The research results indicated that the in-vehicle intelligent agent with a high pitch female voice is most likable and trustworthy among those with low pitch female voice, high pitch male voice, and low pitch male voice. There were no significant differences observed in likeability and trust among last three cases. Moreover, the trust perception of male participants was not influenced by the voice mode of intelligent agents but depended on the characteristics of the autonomous vehicle. The results in the experiment provide some insights of human-computer interaction concerning voice mode for product designers and researchers.