Exploratory Study of Some Acoustic and Articulatory Characteristics of Sad Speech

This study examines acoustic and articulatory EMA data of two female speakers (American and Japanese) spontaneously producing emotional speech while engaged in an informal telephone-type conversation. A set of control data in which the speakers imitated or read the original emotional utterance was also recorded; for the American speaker, the intonation pattern was also imitated. The results suggest (1) acoustic and articulatory characteristics of spontaneous sad speech differ from that of read speech or imitated intonation speech, (2) spontaneous sad speech and imitated sad speech seem to have similar acoustic characteristics (high F(0), changed F(1) as well as voice quality), but articulation is different in terms of lip, jaw and tongue positions, and (3) speech that is rated highly by listeners as sad is associated with high F(0) and changed voice quality.