Table 4 Emotion recognition results based on multi-task learning
From: Speech emotion recognition based on emotion perception
Multi-task learning | UA | WA |
|---|---|---|
A | 68.57% | 67.79% |
B | 68.91% | 67.42% |
C | 68.31% | 67.23% |
D | 67.44% | 66.67% |
A+B | 68.61% | 65.36% |
A+C | 67.46% | 67.23% |
A+D | 67.91% | 64.23% |
B+C | 67.25% | 67.42% |
B+D | 68.75% | 68.16% |
C+D | 68.23% | 68.16% |
A+B+C | 67.84% | 65.92% |
A+B+D | 68.30% | 66.29% |
A+C+D | 68.14% | 66.29% |
B+C+D | 69.38% | 67.04% |
A+B+C+D | 70.42% | 67.79% |
Baseline (single task) | 67.98% | 64.61% |