Subjective and Objective Sound Quality Predictive Models for the Assessment of a Propeller Aircraft Interior Noise
* Presenting author
The interior sound of an aircraft is generally not optimized for passenger acoustic comfort because its assessment often occurs in the late stages of the development cycle. To improve these aspects, the design philosophy should shift to a human-centered paradigm and the study of the sound quality aspects front-loaded. In this paper we discuss a data-driven method for the evaluation of a propeller aircraft interior noise on the basis of objective and subjective psychoacoustic attributes. Such approach, combined with virtual prototyping and sound synthesis tools, paves the way for the inclusion of the human perception in the aircraft design optimization process. The developed instrument grounds on a modular approach capable of classifying in terms of objective sound quality attributes and of subjective passenger annoyance the inputted propeller aircraft in-cabin sound samples, obtained through experimental recordings or through sound synthesis of numerically simulated data. The objective sound quality features are estimated through a set of Convolutional Neural Network models trained on time domain labeled data, while the subjective annoyance is predicted through a feature-based Artificial Neural Network, trained on the basis of a jury test campaign. The paper discusses the accuracy of the method and reports on experimental and numerical applications.