Syllable Segmentation on LibriSpeech (test)
76.6PrecisionSylber
Evaluation Results
| Method | Links | ||||
|---|---|---|---|---|---|
| SylberInitialization=SD-HuBERT, Vocab size=4096, Token frame rate=3.76, w/ distill=true2026.07 | 76.6 | 68.3 | 72.2 | 75.9 | |
| SylReg-DistillInitialization=data2vec 2.0, Vocab size=8192, Token frame rate=5.82, w/ distill=true2026.07 | 64.5 | 88.7 | 74.7 | 63.1 | |
| SD-HuBERTInitialization=HuBERT, Vocab size=4096, Token frame rate=4.67, w/ distill=false2026.07 | 64.3 | 71 | 67.5 | 70.7 | |
| SylBoost 6.25HzInitialization=data2vec 2.0, Vocab size=8192, Token frame rate=5.86, w/ distill=true2026.07 | 64.1 | 88.7 | 74.4 | 62.4 | |
| SylRegInitialization=HuBERT, Vocab size=4096, Token frame rate=6.31, w/ distill=false2026.07 | 60.3 | 89.8 | 72.2 | 54.1 | |
| HuBERTVocab size=4096, Token frame rate=6.71, w/ distill=false2026.07 | 47.9 | 75.9 | 58.7 | 39.1 |