Scripted vs spontaneous speech detection on Internal curated English dataset (val)
97.28AccuracySEAM
Evaluation results
| Method | Links | ||
|---|---|---|---|
| SEAMSeed=1339, Backbone=DistilHuBERT, Window size=8 s, Layer unfreezing=top-1-layer, Training epochs=32026.06 | 97.28 | 98.35 | |
| SEAMSeed=1337, Backbone=DistilHuBERT, Window size=8 s, Layer unfreezing=top-1-layer, Training epochs=32026.06 | 96.9 | 98.2 | |
| SEAMSeed=Mean, Backbone=DistilHuBERT, Window size=8 s, Layer unfreezing=top-1-layer, Training epochs=32026.06 | 96.73 | 98.06 | |
| SEAMSeed=1338, Backbone=DistilHuBERT, Window size=8 s, Layer unfreezing=top-1-layer, Training epochs=32026.06 | 96.02 | 97.62 |