Speaker Diarization on DebateIndonesianLarge (test)
77.78PrecisionIndo Adapted (25h)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Indo Adapted (25h)Evaluation Protocol=Fine-tuned, Training Data Volume=25 hours2026.01 | 77.78 | 99.06 | 87.14 | 29.24 | |
| Indo Adapted (2h)Evaluation Protocol=Fine-tuned, Training Data Volume=2 hours2026.01 | 74.18 | 100 | 85.17 | 34.81 | |
| AMI BaselineBackbone=pyannote/segmentation-3.0, Evaluation Protocol=Zero-shot2026.01 | 68.18 | 87.23 | 76.54 | 53.47 |