This study investigates the potential of sound analysis to detect bee states within beehives, a critical challenge for beekeepers. We propose a novel approach combining Mel-Frequency Cepstral Coefficients (MFCC) and Short-Time Fourier Transform (STFT), effectively creating more informative representative data for the classification of bee states. Experimentation on a real dataset demonstrates that the Random Forest classifier utilizing synergy features extracted from MFCC and STFT outperforms models relying solely on MFCC or STFT features, significantly improving classification accuracy (up to 87.2%). This integrated approach offers advantages in capturing both spectral detail (MFCC) and temporal information (STFT) potentially leading to improved classification accuracy for bee states detection. The findings contribute valuable insights for developing robust bee colony health monitoring systems.