The company Kien 1 is developing Bluetooth Low Energy (BLE) speakers, consisting of a subwoofer and several satellite speakers. These subwoofer and satellite speakers will have a feature named Follow Me, which when enabled from a smartphone application, will play music only in the room where the user currently is. The BLE’s signals will encounter obstruction, reflection, and diffraction from obstacles such as people, walls, and furniture. These indoor propagation issues cause the received signal strength (RSS) to have more distortion and variate more. Indoor position systems (IPS) that make use of a geometric model, such as using the RSS in trilateration methods, will suffer greatly from the propagation issues. On the contrary, fingerprinting methods that use a statistical model are more robust. However, fingerprinting requires a time consuming setup that needs to be redone periodically and every time after the environment changes. We implemented a Naive Bayes Gaussian classifier that can distinguish whether the signals travelling between a pair of BLE speakers, which are placed in two fully furnished adjacent rooms, came from speakers in the same room or in different rooms. Subsequently, the speakers that are in the same room can be grouped together, which is a subtask for the Follow Me feature. The classifier uses two features derived from the RSS information. Namely, the mean of the sum of several RSS samples (MSUM) and the standard deviation of the RSS range of several sample sets (SRR). We have quantified the classifier’s robustness by evaluating its performance in 9 different environments that had obstacles or radio frequency (RF) interference or both. We chose the acceptable performance to be an area under the curve (AUC) of at least 80.00 %. The classifier was able to successfully distinguish between BLE speakers that were in the same room and in different rooms in 7 environments with a minimum and maximum AUC of 86.64 % and 100.00 %, respectively.