Analyzing the Beehive’s Sound to Monitor the Presence of the Queen Bee

In this paper, a method for beehive monitoring is proposed that is focused on the spectral analysis of the bees’ sound. Towards that direction, a system was implemented that allowed for constant acoustic monitoring and the creation of a database of audio recordings. Also, an experiment was performed to simulate two different states of the beehive’s colony. One healthy state with the queen bee present and normal bee activity and one state where the queen bee is removed. The two different states are being analyzed in terms of their spectral content to establish their characteristic frequencies. Also, a set of features are extracted, the Mel Frequency Cepstral Coefficients (MFCC), and their distribution on each mel band is examined for a set of data taken from multiple days on two different beehives. By comparing these distributions, it was possible to make a clear distinction between these two states for both hives and to detect the state of the beehive based on halfhour measurements.