“…Fragopanagos et al [12] state that most research efforts investigated the affective speech processing on the level of complete utterances, words, or phonetic transcription independent chunks [13,14]. A comparably smaller number of methods are based on phonetic pattern modeling within emotion classification [15,16,17,18,19,20,21,22,23,24,25,26]. Several studies already reported accuracies on multiple corpora -however, only a very few consider training on one and testing on a different one (e.g., [27] and [28], where two and four corpora are employed, respectively).…”