These slides were presented at the NAFEMS World Congress 2025, held in Salzburg, Austria from May 19–22, 2025.

Abstract

In the case of NVH performance evaluation, there have been limitations in applying virtual vehicle development so far. The first reason is the technical issue of implementing 3D visualization of noise in a virtual environment. The second reason is the difference between the noise perceived by drivers in a vehicle and the reproduced noise in a virtual environment. As a solution to these issues, recent developments in spatial audio technology and 3D sound color mapping have received attention. By utilizing spatial audio technology, it is possible to provide the same experience to the driver even without physical driving. With the use of visualization technology, it becomes easier to recognize major noise sources and noise paths. In this paper, Ambisonics was used as the 3D sound format for implementing spatial audio and color mapping. Ambisonics is a technology that converts spatial audio information into a three-dimensional sound field composed of directional characteristics based on Spherical Harmonics. It is used for recording or reproducing immersive sound. As the order of the spherical harmonic increases, the number of required listening points increases, but it also allows for more sophisticated sound representation. For recording Ambisonics beyond the 3rd order, it is common to use microphone arrays in a spherical arrangement to capture 3D sound. This allows for the measurement of higher-order Ambisonics. In the case of the Eigenmike-32, where 32 microphone capsules are distributed on a surface of a 42mm radius microphone array, the effective frequency range for applying third-order Ambisonics is limited to 700Hz to 8kHz, which is insufficient to cover the primary frequency range of interest in the development of NVH performance for completed vehicles, which is from 20Hz to 1kHz. In this study, we conducted NVH virtual performance evaluation using a new concept helmet microphone array that is suitable for measuring noise in the frequency range of 20Hz to 1kHz, which is the range of interest for vehicle driving noise. The helmet microphone array has the advantage of being able to measure sound from the driver's seat most realistically during driving, and its larger radius of 95mm allows for a 2.3 times larger size compared to the commonly used Eigenmike-32, making it suitable for measuring low-frequency noise below 1kHz. In this study, we have developed a process to generate spatial audio in Ambisonics and Sound Particle Separation (SPS) format based on sound captured using a helmet microphone array. This process also includes post-processing techniques to make the spatial audio audible and visualizable. The raw data used in this study was acquired through measurements or simulation including a helmet microphone array model, which provided us with the extracted impulse responses. We then applied auditory filters to the extracted data and synthesized them to transform them into spatial audio format. Through this process, we were able to implement auditory rendering and visualization techniques.