Comparison of some listening test methods : a case study
Résumé
The goal of this study was to compare various listening test methods in the particular case of nine in-car ventilation noises. Six listening tests were conducted : - absolute evaluation of noise pleasantness; - evaluation of pleasantness, during which the subject could hear all noises as often as necessary; - paired comparisons (forced choice procedure); - paired comparisons (five levels scale); - paired comparisons (continuous scale); - similarity ratings, analysed with a multidimensional scaling method (Indscal). These six tests were realised by 64 subjects. Different items were examined for each test : its perceived and real duration, its estimated difficulty, the accuracy of merit scores attributed to noises, the perceptual spaces which could be built and the accuracy of a pleasantness indicator computed from the results. It appeared that : - the second procedure can propose a good compromise between the accuracy of the results and the time needed for subjects to realise the test. Thus, it can be recommended for many industrial purposes; - however, the not-forced choice paired comparisons enable a greater discrimination between stimuli; - perceptual spaces built from the paired comparison tests and the similarity rating one were similar, indicating a great stability of sound features used by listeners, whatever their task.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...