To consider the acoustic quality in car design, a scientific methodology is developed in order to give an objective notation. To obtain a good correlation between the subjective and objective judgments, the first point is the choice of recording points that represent the signature of each sound sources that the human ear detects in space. Then, the information about the sound quality is extracted from signals with the computation of different appropriate parameters like “shock balance” and “roughness”. From these metrics, a notation system similar to a neural network gives a single note. At last, the processing of parameter is different for engine in stationary or run-up conditions.