Pronunciation Assessment on Speechocean762 (test)
80.7Utterance Acc (PCC)HMamba
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| HMambaSupervision Type=Supervised2026.06 | 80.7 | — | — | — | — | 84.8 | 84.3 | — | |
| HiPPOcurricular strategy=false, CONO regularizer=false2025.12 | 79.1 | 0.08 | 0.657 | 63 | 0.643 | 84.5 | 83.7 | 81.6 | |
| Hier-LLaMA2025.12 | 78.9 | 0.082 | 0.656 | 62.2 | 0.634 | 84.4 | 83.2 | 81.1 | |
| Parallel-LLaMA2025.12 | 77.4 | 0.074 | 0.658 | 59.8 | 0.61 | 83.7 | 82.9 | 79.6 | |
| 3M2025.12 | 76 | 0.078 | 0.656 | 59.8 | 0.617 | 82.8 | 82.7 | 79.6 | |
| Parallel-TFR2025.12 | 75.4 | 0.078 | 0.65 | 57.5 | 0.589 | 81.6 | 80.6 | 77.2 | |
| GOPTSupervision Type=Supervised2026.06 | 74 | — | — | — | — | — | — | — | |
| HierTFR2025.12 | 73.5 | 0.081 | 0.644 | 62.2 | 0.634 | 80.1 | 79.5 | 76.4 | |
| HiPAMA2025.12 | 73 | 0.084 | 0.616 | 57.5 | 0.591 | 74.9 | 75.1 | 75.4 | |
| DeepFeatureSupervision Type=Supervised2026.06 | 72 | — | — | — | — | — | — | — | |
| GOPT2025.12 | 71.4 | 0.085 | 0.612 | 53.3 | 0.549 | 75.3 | 76 | 74.2 | |
| MultiPASupervision Type=Supervised2026.06 | 70.5 | — | — | — | — | 77.2 | 76.4 | — | |
| AzurePA2025.12 | 70 | — | — | 62.3 | — | 71.5 | 84.2 | 78.2 | |
| Audio + transcript-guidedSupervision Type=Zero-shot / unsupervised, Training Data Volume=100 hours, Input Modality=Audio + transcript-guided2026.06 | 66.8 | — | — | — | — | 75.7 | 74.8 | — | |
| Audio + transcript-guidedSupervision Type=Zero-shot / unsupervised, Training Data Volume=960 hours, Input Modality=Audio + transcript-guided2026.06 | 66.1 | — | — | — | — | 76.3 | 75.3 | — | |
| GoPSupervision Type=Supervised2026.06 | 64 | — | — | — | — | — | — | — | |
| Transcript-only (DTW dist.)Supervision Type=Zero-shot / unsupervised, Training Data Volume=960 hours, Input Modality=Transcript-only2026.06 | 63.3 | — | — | — | — | 70.9 | 70.7 | — | |
| Transcript-only (DTW dist.)Supervision Type=Zero-shot / unsupervised, Training Data Volume=100 hours, Input Modality=Transcript-only2026.06 | 61.1 | — | — | — | — | 66.4 | 66.8 | — | |
| Audio-onlySupervision Type=Zero-shot / unsupervised, Training Data Volume=100 hours, Input Modality=Audio-only2026.06 | 60.1 | — | — | — | — | 69.4 | 68 | — | |
| Liu et al. (aMRT)Supervision Type=Zero-shot / unsupervised2026.06 | 60 | — | — | — | — | — | — | — | |
| Audio-onlySupervision Type=Zero-shot / unsupervised, Training Data Volume=960 hours, Input Modality=Audio-only2026.06 | 59.7 | — | — | — | — | 69.4 | 68.8 | — | |
| non-reg GoPSupervision Type=Zero-shot / unsupervised2026.06 | 57 | — | — | — | — | — | — | — | |
| HiPPOCONO regularization=enabled, Curriculum Learning=enabled2025.12 | 0.733 | 0.202 | 0.48 | 0.52 | 0.521 | 80.6 | 0.797 | 0.754 | |
| Parallel-LLaMA2025.12 | 0.726 | 0.237 | 0.345 | 0.426 | 0.428 | 79.9 | 0.791 | 0.748 | |
| HiPPOCONO regularization=disabled, Curriculum Learning=enabled2025.12 | 0.72 | 0.213 | 0.448 | 0.513 | 0.516 | 79.7 | 0.791 | 0.743 | |
| Parallel-TFR2025.12 | 0.717 | 0.24 | 0.33 | 0.416 | 0.417 | 79.7 | 0.791 | 0.741 | |
| MultiPA2025.12 | 0.705 | — | — | 0.427 | 0.436 | 77.2 | 0.763 | 0.73 | |
| HiPPOCONO regularization=enabled, Curriculum Learning=disabled2025.12 | 0.698 | 0.241 | 0.331 | 0.404 | 0.404 | 79 | 0.785 | 0.728 | |
| VanillaSSL2025.12 | 0.692 | — | — | — | — | 75.7 | 0.757 | 0.714 | |
| Hier-LLaMA2025.12 | 0.692 | 0.238 | 0.328 | 0.412 | 0.418 | 78.6 | 0.78 | 0.724 | |
| Liu20232025.12 | — | — | — | — | — | 79.5 | — | — |