This paper presents a novel approach to design target sounds. The approach combines two interesting and powerful technologies: (1) a fully automatic sound feature extraction algorithm to identify the most disturbing parts in the time-frequency spectrogram of a sound and (2) a model-based sound synthesis approach with virtual modification capabilities. By doing welloriented modifications in the critical time-frequency parts of the sound, engineers can easily resynthesize a new sound with better sound quality. This sound can be used as a target for an intelligent structural redesign of the vehicle or aircraft.
