Search Results

E-vehicles can generate strong tonal components that may disturb people inside the vehicle. However, such components, deliberately generated, may be necessary to meet audibility standards that ensure the safety of pedestrians outside the vehicle. A tradeoff must be made between pedestrian audibility and internal sound quality, but any iteration that requires additional measurements is costly. One solution to this problem is to modify the recorded signals to find the variant with the best sound quality that complies with regulations. This is only possible if there is a good separation of the tonal components of the signal. In this work, a method is proposed that uses the High-resolution Spectral Analysis (HSA) to extract the tonal components of the signal, which can then be recombined to optimize any sound quality metric, such as the tonality using the Sottek Hearing Model (standardized in ECMA 418-2).

To empirically estimate the radiation of sound sources, a measurement with microphone arrays is required. These are used to solve an inverse problem that provides the radiation characteristics of the source. The resolution of this estimation is a function of the number of microphones used and their position due to spatial aliasing. To improve the radiation resolution for the same number of microphones compared to standard methods (Ridge and Lasso), a method based on normalizing flows is proposed that uses neural networks to learn empirical priors from the radiation data. The method then uses these learned priors to regularize the inverse source identification problem. The effects of different microphone arrays on the accuracy of the method is simulated in order to verify how much additional resolution can be obtained with the additional prior information.

Clearly, sound quality evaluation has become a central focus for assuring customer satisfaction. To achieve an optimized product sound at an early stage of development, subjective evaluation methods must be combined with analysis and prediction tools to provide reliable information relevant to product quality judgments. Some years ago, a “Hearing Model” was developed explaining and describing many psychoacoustic effects [1], [2], and allowing for roughness calculation in accordance with subjective listening tests [3]. Existing roughness models work well for synthetic signals such as modulated tones or noise signals, but it is challenging to predict roughness for engine sounds because of their more complex spectral and temporal noise patterns [4].