Acoustic characteristics predicting the annoyance of interior train background noise
Résumé
One the main advantages of railway transportation is that it allows passengers to engage in a variety of activities during train rides: relaxing, reading, working, watching movies, chatting with friends etc. However, much of these activities require a relatively quiet environment. It is therefore utterly important for train operators to ensure that they offer such quiet environments, tailored to its customers? needs. Previous studies have shown that the main sources of annoyance for passengers are other passengers? noises, squeaks and rattles, and train background noise: too loud background noises, with unbalanced spectra, and audible tonal components are annoying and make some activities tiresome. However, current specifications of rolling stock account only partially for passenger perception and miss important annoying acoustic characteristics of background noise. The goal of this work was therefore to investigate more subtle effects of background noise on passengers? perceived annoyance, and provide a global annoyance indicator based on a combination of psychoacoustic descriptors. The study was based on a database of binaural recordings of background noises of train operating at different speeds. During a listening test, participants listened to a selection of these recordings through near-field transaural reproduction. The test used an original MUSHRA-inspired method, in which participants compared different recordings and reported perceived annoyance on a continuous scale. The stimuli were grouped together so as to limit loudness variations that may cancel more subtle acoustic effects. A set of recurring anchors were used for each group of sounds, and the anchors were used to scale the different comparisons. Multilinear regression finally created a global annoyance indicator based on a combination of psychoacoustic indicators.
Origine : Fichiers éditeurs autorisés sur une archive ouverte