AES 5th International Conference on Automotive Audio

The audio processing requirements for automotive systems continue to grow. This is driven by a combination of concurrent use cases and high channel counts. To support this, modern SOCs and DSPs employ multiple threads and cores to increase computational throughput. This paper discusses how to best leverage these hardware features for audio processing work loads and how to accurately profile systems in a multi-threaded environment.

ID:03_Beckmann

This paper presents a methodology for evaluating the perceived quality of an audio processing device. It is based on two stages, namely Objective and Subjective.

The Objective stage consists of producing objective measurements of audio signal descriptors of both source and processed signals. Based on the collected dataset, deviations to Target Reference values are computed and utilized to generate statistical analysis on the performance of the device.
The Subjective stage is performed to assess the quality of the processed audio content as perceived by the end-users. Based on a Mean Opinion Score framework, it collects subjective assessments on the following attributes: Target Consistency and Impact on Source Quality. The results are processed to describe the performance of the device being assessed.

As a sample implementation of our methodology, the performance of three different loudness processors are evaluated and the results of Objective and Subjective stages are used for defining their overall quality performance.

ID:03_Travaglini

While traditional automotive audio systems were used to a variety of architectures, the global trend in automotive is to centralization and unification in the electrical and electronics (E/E) architecture.
In this document, we detail the impact the E/E architecture evolution has on audio processing design and implementation projects, up to the sound tuning in pre-production. We first understand the variety of traditional architectures and their implications, before following the differences observed on newer, centralized architectures from an audio processing development and project points of view.

ID:03_Andriamakaoly

Analysing and simulating loudspeaker systems in vehicles is a complex task. In order to be able to make a statement about the sound characteristics at an early stage of development, various physical disciplines must be coupled together. The simulation starts with electromagnetism and is followed by mechanical modelling, which in turn is linked to the acoustic model via the diaphragm speed. Some audio systems consist of more than 20 loudspeakers that are used simultaneously, each of which is designed for a different frequency range. A rough categorisation is made here into the low, mid and high frequency ranges, whereby each of the ranges presents different difficulties for the simulation.
The focus of this article is on the high frequency range, which places great demands on the acoustic simulation with classical discretisation methods, such as the finite element method (FEM) or the boundary element method (BEM), particularly due to the short wavelengths. In order to obtain the prediction in the entire frequency range up to 20,000 Hz, either accelerated methods, such as the fast multipole BEM, or energetic methods, such as the energy BEM/FEM, must be used. Their feasibility is to be validated using a generic car measurement setup. In addition to the numerical challenge, there are also special demands on the measurement setup in this area, as only small changes in the microphone position can lead to significant differences due to the short wavelengths.

ID:03_Dilba

Personalized sound zones technology is a groundbreaking solution for enhancing the travel experience in consumer vehicles. By creating acoustically separated areas within the car interior, this technology allows for novel use cases, such as robust individual interaction with a voice assistant for the driver and simultaneous high-quality private voice calls for passengers.

Personalized sound zones require two components: source separation at voice capture and personalized audio reproduction. This paper focuses on source separation, by analyzing how microphone types and positioning affect the starting conditions for the task and presenting a solution that improves the separation and enhancement performance for any microphone type.

ID:03_Buccoli

This paper introduces a novel approach to enhancing loudspeaker performance by employing Long Short-Term Memory (LSTM) neural networks to linearize the driving force on the voice coil. The method enables transducer engineers to design more cost-effective drivers by prioritizing the optimization of mechanical components (such as the membrane and suspension), while allowing for greater compromises and flexibility in the design of the magnetic system (including the voice coil and motor), whose deficiencies are compensated by the controller. This flexibility is particularly beneficial in challenging applications where size, cost, and weight constraints are significant factors, such as in automotive and portable devices.

We propose a control algorithm that linearizes the driving force by adjusting the voice coil current and compensating for the force factor BL nonlinearity. By focusing on the current signal, which is readily obtainable, and the BL nonlinearity, which is relatively stable and predictable, our controller architecture remains practical and robust. This intentional avoidance of complex and potentially time-varying mechanical nonlinearities and the associated costly acquisition of mechanical signals (e.g., displacement, velocity, and acceleration) ensures feasibility, especially in production environments.

The performance of the controller was rigorously evaluated using a real 2.5-inch driver deliberately engineered with compromised voice coil characteristics, resulting in a distorted BL curve and heightened levels of distortion. Comparative analysis was conducted against a non-controlled sample of a similar driver with a properly designed magnetic system (i.e., both drivers sharing the same mechanical system, but the controlled unit exhibiting suboptimal magnetic properties). Our proposed controller algorithm achieved approximately 10 dB reduction in distortion above the resonance region, approaching the performance of the well-designed unit.

ID:03_Volkov

Several studies have shown that individual preferences with respect to frequency- and loudness-dependent amplification can differ markedly across listeners. To accommodate such individual preferences, sound personalization concepts have been proposed. This study investigates if differences in individually preferred settings across different audio systems can be predicted from measurable differences in acoustic transmission properties of the audio systems. Self-adjusted preference profiles of 20 normal-hearing listeners were measured for seven audio playback systems including automotive audio systems, a soundbar, and two types of headphones. The results indicate that, while interindividual preferences varied largely, there were systematic differences between audio systems which appeared to compensate acoustic differences, suggesting that future applications could benefit from known individual preference data in other playback systems to automatically adapt to an individual user’s needs.

As immersive object-based spatial audio continues to gain widespread adoption in the music industry, automobiles have become a target environment for multichannel audio reproduction capabilities. This emergence is driving new advances in audio rendering techniques. We present novel rendering strategies that scale with the number of available channels in automotive audio systems. This includes virtual height reproduction without the need for discrete overhead speakers that can be costly to implement. Here we compare the perceptual impact of 5.1 discrete, 5.1.4 discrete, and 5.1 virtual height rendering systems in multiple automotive environments. Subjective assessment paired comparison experiments were conducted to measure preference and multiple spatial audio attributes, and probability outcomes were modelled using a Bradley-Terry-Luce rank ordering. The 5.1 virtual height system provides a comparable sense of spaciousness and externalization to the 5.1.4 discrete channel system without degradation of spectral naturalness. These findings highlight the potential for virtual height rendering to enhance spatial audio immersion in automotive environments while minimizing the need for additional physical speakers. Further, this methodology can be generalized for subjective evaluation of audio delivery performance across a variety of channel configurations. We also describe multiple alternative rendering techniques for virtualization of immersive audio in automotive environments for future investigation.

Automotive manufacturers are engaging with audio system providers to offer unique auditory experiences. This is achieved through the replication of acoustic environments, coinciding with the growing prominence of infotainment systems. Gaining an understanding of how acoustical parameters impact human perception is of paramount importance in the development of virtual acoustic venues within this context. Hence, this is the initial study to explore the correlation between objective measures and subjective responses in the reproduced in-vehicle virtual acoustics, with the aim of augmenting the auditory experience for passengers. A jury test was conducted, and the resulting data was subjected to statistical analysis. The findings align with previous studies except for intimacy, indicating the visual disparities in virtual environments. Also, while reverberation, closely linked to envelopment, is basically influenced by early decay time, optimizing reverberation time within a specific range can improve the auditory experience in vehicles by naturality, which is correlated with key variables including overall impression.

The loudspeaker configuration in a car is far from a conventional stereophonic one. The aim of this study was to evaluate the consequences of such an unusual configuration on the perception of the localisation of virtual sound sources. Experiments on the directional perception of virtual sources in an automotive environment were performed using conventional loudspeaker configurations. The sources, consisting of broadband pink noise bursts, were laterally spatialised using the Virtual Based Amplitude Panning (VBAP) method. Participants were asked to point their nose in the direction of the perceived sound source. Their head movements were recorded by a headtracker. The differences between the perceived directions and those predicted by the VBAP method in a classical stereo configuration were measured. In addition, several models of azimuth perception were compared with the experimental data. An extended energy vector model, taking into account the precedence effect, gave results in agreement with the experimental data. The model inputs are the relative position, time delay and level of each loudspeaker. This model could be used as a tool for assessing spatial restitution, loudspeaker implementation and audio tuning.

This paper investigates the perceived spatial extent of a local active noise control (ANC) system for different types of disturbances. Several publications have determined size and shape of the zone of quiet for various arrangements, primarily for pure-tone diffuse sound fields, through analytical and numerical methods. However, these studies have often overlooked human perception, focusing solely on technical properties. Therefore, a listening experiment has been conducted to determine the perceived size of the zone of comfort in a scenario close to reality, using an active headrest setup. Several operational frequency limits for different types and directions of broadband disturbances are examined. Within this experiment, lateral transitions to the front and head rotations at the target position have been considered. Statistically consolidated subjective ratings exhibit limits of around 2 cm to 4 cm for lateral transitions, with an expected decrease towards higher frequencies. When comparing participants’ answers to measurements, the majority of the median responses converge at a point with loudness reduction of 20%. The rotational limits of 7° to 15° are not as dependent on frequency, but rather on the distinct perception of the secondary sources and the loudness reduction as well.

This paper presents a synthesis of research spanning eight decades, investigating the relationship between auditory spatial perception and the effects of masking noise. Drawing from studies in psychoacoustics, it examines how different characteristics of masking noise affect fundamental auditory experiences such as sound localization, echo suppression, precedence effect, and similar phenomena.

Despite the extensive research, the practical application of these findings in audio products engineering (such as automotive sound systems) remains limited. Incorporating these scholarly insights into practical use could improve algorithms for spatial sound processing. Understanding the implications outlined in these studies is crucial for refining immersive sound reproduction and implementing new spatial audio formats into consumer devices.

The challenges posed by noise, especially in vehicles, emphasize the importance of applying prior research to enhance the user experience. This review highlights the need for understanding and applying insights from previous studies to improve the quality of experience of consumer audio devices and systems.

This paper explores the use of virtual sensing in a feedforward active noise cancellation (ANC) system for an electric vehicle. Vibration reference signals and microphone disturbance signals were recorded in an electric vehicle operating in cruise control mode at various driving speeds. The dataset was used to test the Remote Microphone (RM) and Auxiliary Filter (AF) virtual sensing methods and to investigate their sensitivity to variations in road speed and acoustic conditions. The standard RM method gave slightly better results than the AF method and was less computationally expensive. This was improved upon using a variant of the standard RM that minimises a delayed version of the error. The optimisation methods and control algorithms developed here could be readily used on other datasets to further investigate the use of virtual sensing.

Haptic technologies are being increasingly employed as part of automotive sound reproduction systems to enhance listener immersion within the rendered soundfield. Recent research on the use of audio haptic transducers as a means of providing a multi-sensory experience of sound has focused on characterising the performance of transduction devices. To date, performance evaluation has been restricted to experimental conditions in which haptic transducers have been measured individually with specific unit mounting and load conditions. This research demonstrates the validity of extending existing performance evaluation methods and metrics to in situ systems (e.g. those employed in an automotive setting). Specifically, a system of four individual haptic transducers mounted on the backrest of an automotive seat is evaluated, with and without system tuning strategies applied, using accelerometer measurements. Results demonstrate that system response characteristics can be tuned via frequency equalization and input signal dynamics pre-processing with the tuning quantified through use of existing performance metrics. In particular, it is shown that the overall response of the system under test may be optimised through analysis and tuning of each separate transducer individually.

Measuring in-car audio performance is an important task as it allows car makers and Tier 1 suppliers to objectively validate, evaluate and compare audio systems across different car models. To address the lack of consensus on measures and procedures, the Technical Committee of Automotive Audio of the Audio Engineering Society created a white paper that draws a standard procedure for in-car acoustic measurements. This article presents a comprehensive study conducted following the guidelines of the white paper on automotive audio measurements. The study was carried out on two models, including a high-performance super car, providing a unique perspective on the applicability of these guidelines in the domain of sports cars. This study aims at contributing to the ongoing discourse in automotive audio engineering, emphasizing the need for adaptable and flexible measurement standards for the widest range of vehicle types.