The acoustic design of concert halls, opera houses, arenas and auditoria is usually based on the study of some objective acoustic parameters determined on the basis of the computed or measured impulse response. Such objective parameters are based on the average response of the human ear to transient sounds, without considering the subjective perception of the individual, which is in turn related also to social ad education aspects. The aim of this study is to investigate the relationships between individual sound perceptions and energy parameters. In detail, a questionnaire has been submitted to a sample of people hearing different tracks pertaining to different musical genres. The questionnaire has been conceived and scrutinized to come up with a classification model for the audience allowing to express the listener's experience according to different parameters: overall acoustic quality, sharpness of articulation of the melodic sequence, reverberation, winding, intensity, background noise, sharp balance for mid frequencies, tonal balance and overall acoustic experience. Results of this experiment are analysed by means of non parametric models belonging to the data mining algorithms such as Random Forest (Breiman, 2001) and Regression Tree (Breiman et al. 1984). The final purpose of the study is to define which physical and acoustical characteristics are the most suitable to represent a specific sound field, responsible for the subjective perceptions of the listeners inside a concert hall, a theatre, an arena or a stadium. Moreover, by means of the statistical approach used to analysed the questionnaire items, the feeling generated by the music on the audience during a concert is evaluated.
