Random Forest-Powered Acoustic Gunshot Detection Leveraging LPC Features and LIME-Based Interpretability

Defense, forensic science, law enforcement, and public safety are the main domains where, gunshot detection technology finds its applications. Rapid detection is crucial for prompt response and damage mitigation. This study utilizes Mel-frequency cepstral coefficient (MFCC) and linear predictive coding (LPC), sound features to accurately identify the gunshot sounds from the audio inputs. The results show that the Random Forest classifier, when combined with LPC features, achieves the near-perfect test set accuracy, that is close to 100%. One of the key aspects of this research is the integration of LIME-based analysis provides valuable insights into the model’s decision-making process, it showcases that it is realiable and highlights the significant role of the LPC features in gunshot detection.