Scott Amman

Guidelines for Jury Evaluations of Automotive Sounds

This article provides a set of guidelines intended to be used as a reference for the practicing automotive sound quality (SQ) engineer with the potential for application to the field of general consumer product sound quality. Practicing automotive sound quality engineers are those individuals responsible for understanding and/or conducting the physical and perceptual measurement of automotive sound. This material draws upon the experience of the four authors and thus contains a number of rules-of-thumb which the authors have found worked well in their many automotive related sound quality projects over past years. When necessary, more detailed publications are referenced. The intent here is to provide a reference to assist in automotive sound quality work efforts and to solicit feedback from the general sound quality community as to the completeness of the material presented. Why is there subjective testing and analysis in automotive sound quality investigations? One might ask why bother with the trouble of conducting subjective testing in the first place? In the authors’ experience, conducting subjective jury evaluations of automotive sounds has led to a deeper understanding of those sounds and the way potential customers react to and sometimes appreciate automotive sounds. The following is an attempt to describe subjective testing and analysis as applied to sound quality and its relevance to gaining this deeper understanding. The remainder of this article draws upon the experience of the four authors and as a result, may be biased toward the techniques they have commonly used or have found to work well in their automotive sound quality studies. However, an attempt has been made to address other techniques commonly used by other researchers in the general field of product sound quality. Although not a comprehensive document, it is hoped that this article will provide a set of guidelines which addresses a majority of the issues and techniques used in the field of automotive and general product sound quality. It is hoped that this guide will act as a springboard: a launching point for your own individual investigation into subjective testing and analysis for automotive sound quality.

An efficient technique for modeling and synthesis of automotive engine sounds

This paper presents a new method for modeling and synthesis of automotive engine sounds using a deterministic-stochastic signal decomposition approach. First, the deterministic component is extracted using a synchronous discrete Fourier transform method and this is subtracted out from the original signal. Next, the (residual) stochastic component is modeled (and synthesized) using a new multipulse excited time-series modeling technique. The effectiveness of the proposed methodology is demonstrated using recorded data sets of actual engine sounds. The results of both numerical and subjective assessment tests are presented.

Sound and vibration perceptual contributions during vehicle transient and steady-state road inputs

This paper discusses two studies which address the perception of both transient and steady state sound and vibration experienced by drivers in vehicles. While past studies with railway and aircraft sound and vibration showed varying degrees of sound and vibration interaction, the results of these two studies showed little evidence of interaction. Additionally, both sound and vibration appeared to contribute about equally to the overall NVH perception. Other researchers have observed that the strongest modality tends to dominate perception, but that effect was not observed in the studies presented here.

On using heterogeneous data for vehicle-based speech recognition: A DNN-based approach

Most automatic speech recognition (ASR) systems incorporate a single source of information about their input, namely, features and transformations derived from the speech signal. However, in many applications, e.g., vehicle-based speech recognition, sensor data and environmental information are often available to complement audio information. In this paper, we show how these data can be used to improve hybrid DNN-HMM ASR systems for a vehicle-based speech recognition task. Feature fusion is accomplished by augmenting acoustic features with additional side information before being presented to the DNN acoustic model. The additional features are extracted from the vehicle speed, HVAC status, windshield wiper status, and vehicle type. This supplementary information improves the DNNs ability to discriminate phonetic events in an environment-aware way without having to make any modification to the DNN training algorithms. Experimental results show that heterogeneous data are effective irrespective of whether cross-entropy or sequence training is used. For CE training, a WER reduction of 6.3% is obtained, while sequential training reduces it by 5.5%.

Sound Quality Metric Development and Application for Impulsive Engine Noise

Many engine tick and knock issues are clearly audible, yet cannot be characterized by common sound quality metrics such as time-varying loudness, sharpness, fluctuation strength, or roughness. This paper summarizes the recent development and application of an objective metric that agrees with subjective impressions of impulsive engine noise. The metric is based on a general impulsive noise model [1], consisting of a psychoacoustic processing stage followed by a transient detection stage. The psychoacoustic stage is extracted from portions of a time-varying loudness model. The primary output of the impulsive engine noise model is a time series that indicates the location and “intensity” of impulsive engine noise events. The information in this time series is reduced either to a single number metric, or to a frequency-based vector of numbers that indicates the amount of impulsiveness in the recorded sound. The frequency-based vector is a distribution of the impulsive engine noise metric as a function of frequency. This is a new development of the impulsive noise model and has proved useful in many applications to indicate which frequency bands contribute most to the overall metric value. An overview of the model and application to various vehicle-level powertrain sounds are presented in this report. Results show distribution of the impulsive engine noise metric as a function of frequency and its agreement with subjective assessments is discussed.

Quantifying customer perceptions of impulsive vehicle noise

Sounds such as spark and diesel knock, squeaks and rattles due to body and suspension components, gear rattle, and other impulsive events, can occur in vehicles and be a major source of customer dissatisfaction. It is desirable to not only know the detection thresholds of these impulsive events, but also the relative annoyance they impart on the customer once they become audible. This work describes research addressing both aspects. The first part of the paper presents a generalized detection model of impulsive events in common vehicle background noises (e.g., wind, road, and powertrain noise). Important properties of the model are the combination of impulsive event information across frequency and also the effect of overall background noise level on detection thresholds. Application of the model to predict detection thresholds for spark knock and also squeaks and rattles is presented. Comparisons are made to measured subjective thresholds. The second part of the paper presents research in objectively qua...

Method for predicting human evaluation of latching and unlatching sounds

A analytical method is provided for predicting human evaluation response to the quality of sounds produced by the operation of the power door lock actuator assembly involving the computation of predicted sound preference from recorded data on certain sound parameters of a sample or samples of the power door lock actuator mechanism.

An auditory model for the prediction of detection thresholds of impulsive sound events

Impulsive sound events such as squeaks and rattles experienced in a vehicle can be a major source of customer dissatisfaction. These events almost always occur in the presence of some sort of background noise (e.g., wind, road and powertrain noise). When complete elimination of the impulsive sound event is either not possible, or too costly, the same effect can be had by pushing the level of the sound below the detection threshold. This paper describes the development of an auditory model that has the ability to predict detection thresholds of impulsive sound events in the presence of noise. The model was initially developed using equal‐energy exponentially damped sinusoids with center frequencies of 250, 500, 1000, 2000, 4000 and 8000 Hz mixed with pink noise. It was then validated using four different squeak and rattle sounds each mixed with three background noises recorded in a vehicle (wind, smooth and rough road noise). An up–down Levitt procedure was used for threshold determination. Two subjects pa...

A Simulation Framework for Subjective Listening Evaluation of Synthesized Automotive Sounds

This paper describes a simulation experiment to study the subjective listening evaluation of synthesized automotive sounds. Three types of digitized automotive sounds are used in the experiment: stationary, nonstationary, and impulsive. These are modeled and synthesized using a newly developed multipulse excitation technique, the characteristics of which can be controlled through careful selection of four key parameters. A blind subjective assessment scheme is devised and nine evaluators are asked to judge the qualities of the sounds syn thesized under varying parametric settings. An analysis of variance (ANOVA) performed on the experimental data reveals a number of useful relationships among the (perceived) qual ity, compression efficiency, and parametric set tings. The paper includes a detailed description of the simulation experiment and the results.

Internal combustion engine noise analysis and production using a novel multipulse excited time series modeling technique.

Automobile manufacturers have learned that customers have well‐defined notions as to what an automobile engine should sound like for the various vehicle segments (luxury, sportscar, economy, etc.). Therefore, it is advantageous to model and synthesize various types of engine sounds for customer evaluation. Much work has been done in the area of speech modeling and synthesis based on physical assumptions of human speech production. The intent of this paper is twofold. First discussed are the similarities and differences between speech and engine noise modeling and synthesis problems. Second, application of techniques learned for speech synthesis to that of automotive noise production is demonstrated. Early models of speech production used periodic pulse excitation in a linear predictive coding (LPC) framework. This technique is used to synthesize engine noise. It is then compared to that of a new suboptimal multipulse method proposed by the authors. It is hoped that this paper will pave the path for a bett...