Low noise asphalt pavement mitigates the impact of traffic noise on the ecological environment and surrounding population to a certain extent. To explore the noise reduction characteristics of different asphalt pavements, an indoor simulation device for pavement noise reduction is developed in this paper, based on the principle of pavement noise reduction. The device simulates the process of noise reflection and interference on the surface and inside of the pavement, which consumes the sound energy to achieve the noise reduction effect. The mechanical performance of six different types of asphalt mixtures is first tested. The noise reduction performance of the asphalt mixtures is then studied using the self-made noise testing device and Kundt’s tube method to verify the accuracy of the device. The influence of surface texture on pavement noise is also analyzed, and test results show that the six types of asphalt mixtures used in this study all meet mechanical performance requirements. A strong correlation is also observed between the noise testing device and the traditional Kundt’s tube method. The noise testing device is found to accurately measure the noise reduction characteristics of asphalt mixtures. For noise of different sound pressure levels, the same type of asphalt mixture displays different noise reduction capabilities. The better the noise reduction performance, the lower the sensitivity to sound wave changes, and the more stable the noise reduction effect. Excessive air void content does not significantly improve the pavement noise reduction performance, and it is recommended that the pavement air void content is maintained within the range of 17%–24%. The density of the asphalt mixture is found to be negatively correlated with the noise reduction performance, while the air void content and the mean texture depth is positively correlated with noise reduction. The results of grey relational analysis show that air void content is the most important factor affecting the noise reduction characteristics of asphalt pavement, followed by mean texture depth. Aggregate particle size is also found to exert a significant influence on noise reduction characteristics, and the influence degree of 9.5 mm particles is found to be the greatest, with 1.18 mm and 2.36 mm particles also showing a high degree of influence. Reasonable control of aggregate particle size should therefore be carried out in the design of low noise asphalt pavement.