Phone Alignment on TIMIT
11.26Mean Alignment Error (ms)MAUS
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| MAUSTraining regime=Baseline2026.06 | 11.26 | 63.55 | 86.76 | 97.82 | |
| MAPSNeural aligner=true, Training regime=Baseline2026.06 | 11.46 | 67.86 | 86.74 | 97.12 | |
| MFA 3.0 MFA trainedVersion=3.0, Phone set=MFA, Training regime=Trained on dataset2026.06 | 11.85 | 62.04 | 85.34 | 97.43 | |
| MFA 3.0 ARPA trained +rulesVersion=3.0, Phone set=ARPA, Training regime=Trained on dataset, Phonological rules=true2026.06 | 11.93 | 62.58 | 85.3 | 97.28 | |
| MFA 3.0 ARPA trainedVersion=3.0, Phone set=ARPA, Training regime=Trained on dataset2026.06 | 11.95 | 63.61 | 85.26 | 97.12 | |
| MFA 3.0 MFA trained -PPVersion=3.0, Phone set=MFA, Training regime=Trained on dataset, Pronunciation probability=false2026.06 | 11.95 | 61.68 | 85.45 | 97.35 | |
| MFA 3.0 ARPAVersion=3.0, Phone set=ARPA, Training regime=Pretrained, Adaptation=false2026.06 | 12.11 | 61.85 | 83.56 | 97.4 | |
| MFA 3.0 ARPA +adaptedVersion=3.0, Phone set=ARPA, Training regime=Pretrained, Adaptation=true2026.06 | 12.21 | 61.7 | 83.42 | 97.26 | |
| MFA 3.0 GlobalVersion=3.0, Phone set=Global, Training regime=Pretrained, Adaptation=false2026.06 | 12.36 | 64.08 | 83.11 | 96.8 | |
| MFA 3.0 Global +adaptedVersion=3.0, Phone set=Global, Training regime=Pretrained, Adaptation=true2026.06 | 12.47 | 63.94 | 83.05 | 96.66 | |
| MFA 3.0 MFA trained +rulesVersion=3.0, Phone set=MFA, Training regime=Trained on dataset, Phonological rules=true2026.06 | 13.09 | 61.64 | 83.99 | 96.34 | |
| MFA 3.0 ARPA trained -PPVersion=3.0, Phone set=ARPA, Training regime=Trained on dataset, Pronunciation probability=false2026.06 | 14.02 | 61.89 | 83.23 | 95.11 | |
| MFA ARPA 1.0Version=1.0, Phone set=ARPA, Training regime=Baseline2026.06 | 16.38 | 48.96 | 76.22 | 95.02 | |
| CharsiuNeural aligner=true, Training regime=Baseline2026.06 | 17.79 | 40.96 | 69.73 | 95.46 | |
| SPPASTraining regime=Baseline2026.06 | 21.44 | 35.29 | 64.31 | 92.25 | |
| BFANeural aligner=true, Training regime=Baseline2026.06 | 43.63 | 14.86 | 28.83 | 64.71 |