Toward a Realistic Encoding Model of Auditory Affective Understanding in the Brain