<p>With the rapid development of modern science and technology, the science popularization exhibition hall, as an important public education space, has gradually become an important place to enhance the public’s scientific literacy. Many exhibition halls have shortcomings in the optimization of thermal energy environment, which affects the visitors’ visiting experience and learning efficiency. The purpose of this study is to optimize the thermal environment of the science popularization exhibition hall by using environmental perception and speech recognition technology, improve the interaction of the exhibition and the sense of participation of the audience, and thus enhance the effect of science popularization education. In this paper, sensors are arranged to collect environmental data such as temperature, humidity and light in the exhibition hall, and an environmental perception system is constructed. At the same time, voice recognition technology is used to enable the audience to interact with the exhibits through voice commands. During the study, a series of experiments were designed to evaluate the effects of different thermal environments on audience attention and information absorption. The experimental results show that the optimized thermal environment significantly improves the audience’s comfort and participation enthusiasm. At the same time, the introduction of voice interaction has significantly improved the audience’s learning effect in the exhibition hall, and the knowledge mastery rate has been significantly improved. Therefore, based on the technical means of environmental perception and speech recognition, the thermal energy environment of the science popularization exhibition hall is effectively optimized, and the overall experience and scientific cognition level of the audience are improved.</p>