CORRELATIONS BETWEEN TIME-FREQUENCY SPECTRAL CHARACTERISTICS AND PSYCHOACOUSTIC METRICS PACS REFERENCE: 43.60.c Piñero, Gema; De Diego, María; González, Alberto 1 Dept. Comunicaciones (Universidad Politécnica de Valencia) Camino de Vera s/n 46022 Valencia - Spain Tel: 34-96-3879761 Fax: 34-96-3877309 E-mail: gpinyero@dcom.upv.es ABSTRACT A novel approach to the widespread problem of evaluating the noise quality of a car engine is presented. Traditionally some psychoacoustic metrics of the sound are calculated and used to generate a certain Noise Quality Index (NQI) according to a previous jury assessment. In this paper we relate well-known psychoacoustic metrics - which usually contribute to generate the NQI - to time-frequency characteristics of car noise, in order to find relations between both kinds of parameters. These relations could help engine designers to predict the noise quality just analysing the spectral changes of the noise waveform. Some promising results are showed. 1. INTRODUCTION Sound quality is a term describing an objective measure of the subjective perception to a radiated sound. Knowledge about how people perceive sounds has been applied to engine noise, and more precisely, to engine idling noise (see for example [1-3]), since this noise can have a considerable influence on potential buyers. Nevertheless, to quantify how the individual signal attributes are perceived and combined to give a listener an overall impression of engine noise is still a difficult task. On the other hand, time-frequency (TF) techniques as Wavelets and Wigner-Ville transformation have also been applied to analyse and study the combustion process of different kinds of engines [4-6], showing that TF analysis represents a powerful tool in order to help engine designers to understand the combustion process, which is supposed to be the main source of engine noise. Concerning our problem of noise quality evaluation, usually a wide set of different metrics of the signal are calculated, involving from psychoacoustic features to statistical and spectral parameters [3]. Once the engine noise has been evaluated by a jury and an assessment of the signals has been obtained, the calculated metrics are compared to it. Those metrics showing sufficient correlation with the results of the jury test are selected in order to build an objective measure of the annoyance caused by the engine [2,4]. Some examples of typical metrics used in engine noise are: ISO532B loudness, sharpness, fluctuation strength, roughness, 1 This work has been supported by CICYT grants TIC1999-0444-C02-01 and TIC2000-1683-C03-01.