Global warming has caused an increase for more energy efficient combustion engines. Measuring the energy performance at real time may require many sensors that increase the final cost of the energy system. This paper describes the feasibility of using deep learning Artificial Intelligence (A.I.) methods to estimate energy system performance using acoustical signals. First, an audio recorder was set up to measure the acoustic signals, while taking direct measurements of an aircraft propulsion system. Then, an energy balance equation for the aircraft was calibrated, and transformed into an algorithm that calculates the Specific Total Energy (STE) in real-time by using the direct measurements recorded. The acoustic signatures were filtered out and their statistical features were used to train and test an artificial neural network that outputs the aircraft’s energy state. This process showed that it is possible to create and train models with an R2 as high as 0.99854, while avoiding overfitting; proving that it is feasible to monitor an energy system performance by using acoustic signals.