“…This study takes nine datasets for analysis, which all rely on ratings of valence and arousal, and which, almost without exception, use only real music examples of a music (listed in Table 1). Three of these feature classical music (Dibben, 2004;Gomez & Danuser, 2004;Bigand, Vieillard, Madurell, Marozeau, & Dacquet, 2005), and will be referred to hereafter as studies C1, C2 and C3; two studies (F1 & F2) were based on film soundtracks (Gosselin et al, 2005;; two others (P1 & P2) investigate popular music, (Yang et al, 2007;Schuller, Dorfner, & Rigoll, 2010); and, in order to have a comparison with genre-specific collections, two datasets (M1 & M2) were chosen because they had a mixture of genres (Nater, Abbruzzese, Krebs, & Ehlert, 2006;Wu & Jeng, 2006).…”