In this paper, the statistical properties of near-ultrasound acoustic noise in typical indoor environments are investigated. The results indicate that near-ultrasound noise exhibits impulsive behavior and that its first-order probability density functions can be accurately modeled by using a Gaussian mixture or symmetric alpha-stable (SαS) distributions. Furthermore, a practical approach for estimating the parameters of the SαS noise in near-ultrasound aerial acoustic communication receivers is proposed. The problem of modeling memory effects in near-ultrasound acoustic noise is also considered and a simple noise model with memory is proposed based on multiple linear stable processes. The proposed statistical noise models can be used to analyze, simulate, and optimize the performance of various near-ultrasound aerial acoustic communication systems. It is believed that the analysis may contribute to the development of more efficient near-ultrasound aerial communication systems for a range of practical applications.