The purpose of our study was to determine the criteria for discriminating between ten emotions (surprise, excitement, joy, contentment, relaxedness, drowsiness, sadness, cold-anger, hot-anger, and fear). Using these subscales we calculated the hit-rate of discriminating between the emotions based on a listening task with a forced-choice evaluation. In the first experiment, participants ( n =39) evaluated a Voice Quality Scale for Emotion Evaluation (VQSEE) (Ikemoto & Suzuki, 2008) after listening to vowels expressing the ten emotions. As a result, we determined three significant canonical discriminant functions (warmth, strength, and dullness) with an average hit-rate for the ten emotions of 23% for the canonical discriminant analysis. The average for the forced-choice listening task was 32%. In the second experiment, the participants ( n =22) made their evaluation after listening to sentences expressing the ten emotions. The criteria were the same as for the first experiment. The average hit-rates of the canonical discriminant analysis and the forced-choice evaluation were 35% and 46% respectively.