Despite the prevalence of computer-generated speech, few studies have investigated the direct relationship between an agent’s voice and students’ perception or recall performance. This study investigated the effects of voice (without visual information) on students’ perception ratings and recall performance. Our results indicated that in the absence of visual information, students greatly preferred the human voice. The recall performance, however, indicated that the synthesized voices led to better recall performance. Implications for pedagogical practices are described in the discussion section.