Abstract
Artificial intelligence (AI) used in intelligence, surveillance, and reconnaissance (ISR) has a high application interest in project management. This article presents a result of military research on ISR applicable to monitoring and recognition of audio signals for environmental protection and critical infrastructure. Two databases with environmental sounds were built from open access platforms for training, validation, and testing. The identification characteristics for IA are extracted from the preprocessing of the sounds, obtaining the Mel Frequency Cepstral Coefficients (MFCC). As a result, model performance for more realistic soundstages shows higher accuracy compared to training categories in identifying signal frequency and duration settings. It is concluded that the model is applicable to various environmental scenarios as a low-cost alternative technology to be applied in sustainable project management.
Publisher
Universidad Nacional de Colombia