Drowning of infants and young children, and the unauthorized use of private pools, are among the phenomena that are becoming commonplace nowadays. The existing solutions to these problems are the hiring of human guard or video analytics, which are costly and not always reliable. In this project, we tried a different approach to solve this problem: detecting pool entry by an acoustic sensor. This is a binary classification problem that can be solved using signal processing and machine learning techniques. In addition, we used various algorithms for audio augmentation and for feature selection. Our results show that entry to a pool can be detected by audio recordings, but the data currently available is not sufficient for solving the problem in a robust and generic way.