Speech Emotion Recognition on IEMOCAP (10-fold cross-validation)
70.3Weighted Accuracy (WA)CNN+LSTM Architecture
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| CNN+LSTM ArchitectureFold=3, Session=2, Gender=F2018.02 | 70.3 | 71.3 | |
| CNN+LSTM ArchitectureFold=2, Session=1, Gender=M2018.02 | 68.8 | 67.7 | |
| CNN+LSTM ArchitectureFold=7, Session=4, Gender=F2018.02 | 68.5 | 59.7 | |
| Satt et al.Evaluation Protocol=5 fold cross-valid.2018.02 | 67.3 | 62 | |
| CNN+LSTM ArchitectureEvaluation Protocol=5 best folds2018.02 | 66.9 | 65.3 | |
| CNN+LSTM ArchitectureFold=6, Session=3, Gender=M2018.02 | 66.4 | 56 | |
| CNN+LSTM ArchitectureFold=5, Session=3, Gender=F2018.02 | 64.8 | 52.1 | |
| CNN+LSTM ArchitectureFold=9, Session=5, Gender=F2018.02 | 64.8 | 64.2 | |
| CNN+LSTM ArchitectureEvaluation Protocol=10 fold cross-valid.2018.02 | 64.5 | 61.7 | |
| CNN+LSTM ArchitectureFold=8, Session=4, Gender=M2018.02 | 64.3 | 67.3 | |
| CNN+LSTM ArchitectureFold=1, Session=1, Gender=F2018.02 | 64.1 | 66.4 | |
| Lee and TashevEvaluation Protocol=5 fold cross-valid.2018.02 | 62.9 | 63.9 | |
| CNN+LSTM ArchitectureFold=4, Session=2, Gender=M2018.02 | 62 | 67.6 | |
| CNN+LSTM ArchitectureFold=10, Session=5, Gender=M2018.02 | 51 | 44.2 |