Determination of sound quality of HVAC systems on trains using multivariate analysis

Abstract

An attempt was made to develop a model describing annoyance of noises generated by HVAC (Heating Ventilation Air Conditioning) systems on trains. The work was performed in three sequential parts. The first part contained the recording and reproduction of the HVAC noises. A binaural recording technique was applied and headphones were used for the reproduction. The second part of the study included both subjective and objective evaluations of the recorded noises. In the subjective evaluation 33 subjects judged nine different noise stimuli on annoyance scales. In the objective evaluation several sound quality descriptors and the third-octave-band levels were calculated. In the final part of the study, multivariate analysis was applied to modeling annoyance of the HVAC noises on the basis of the sound quality descriptors and third-octave-band levels. An annoyance prediction model was developed based on sharpness, a sound quality descriptor that describes the high frequency content in the sound, and a newly developed descriptor, tonal ratio. The new descriptor identifies the tonal components by comparing the sound pressure level between 125 and 800 Hz with the two third-octave-band levels adjacent to it. The interpretation of the results is that high frequency content and tonal components are the main contributors to the annoyance of HVAC noises and is where the main focus should be addressed in order to improve the sound quality of future HVAC systems.