Recordings of 6 sopranos were made singing the sustained vowel /ah/ at a com- fortable pitch with four emotions: angry, happy, sad, and neutral. Perception test results with Japanese university students suggest that there was not always a "match" between what the singers intended to sing and what the listeners perceived, and that almost half of the /ah/ were perceived by listeners as sad. Acousitc analysis of the perceived emotions suggests that F0, formant frequen- cies, formant bandwidth, jitter, shimmer, and vibrato contribute signi 12;cantly to how listeners perceive the emotional singing.