HuBERT-B-LS960 (e2e approach)

Reported on 4 benchmarks across 2 tasks · 1 paper

Note: results are matched by exact model name. Different papers may use the same name for different model variants.

Natural Language Processing4 results

Sentiment AnalysisonSLUE
F1 (%)· 2021-11-19
48
best: 63.3 (W2V2-L-LL60K (pipeline approach, uses LM))
SLUE: New Benchmark Tasks for Spoken Language Understanding Evaluation on Natural Speech arXiv:2111.10367
Sentiment AnalysisonSLUE
Recall (%) · 2021-11-19
47.5
best: 60.4 (W2V2-L-LL60K (pipeline approach, uses LM))
SLUE: New Benchmark Tasks for Spoken Language Understanding Evaluation on Natural Speech arXiv:2111.10367
Named Entity Recognition (NER)onSLUE
F1 (%)· 2021-11-19
49.8
best: 69.6 (W2V2-L-LL60K (pipeline approach, uses LM))
SLUE: New Benchmark Tasks for Spoken Language Understanding Evaluation on Natural Speech arXiv:2111.10367
Named Entity Recognition (NER)onSLUE
label-F1 (%)· 2021-11-19
62.9
best: 82.2 (W2V2-L-LL60K (pipeline approach, uses LM))
SLUE: New Benchmark Tasks for Spoken Language Understanding Evaluation on Natural Speech arXiv:2111.10367