List your company

Speech Recognition on LibriSpeech (dev)

1.29WERFAdam
1.17761.93632.6953.4537Oct 20, 2020Oct 1, 2021Sep 13, 2022Aug 26, 2023Aug 7, 2024Jul 20, 2025Jul 2, 2026

Evaluation results

MethodLinks
FAdamArchitecture=Conformer (600M), Pre-training=w2v-BERT, Fine-tuning protocol=Semi-supervised (LibriLight), Optimizer=FAdam2024.051.29——
Adam (w2v-BERT paper)Architecture=Conformer (600M), Pre-training=w2v-BERT, Optimizer=Adam, Source=[53]2024.051.3——
AdamArchitecture=Conformer (600M), Pre-training=w2v-BERT, Fine-tuning protocol=Semi-supervised (LibriLight), Optimizer=Adam2024.051.3——
Gen3 Conformer XXLUnlabeled data (hrs)=60k, LM Fusion=No2020.101.3——
Gen3 Conformer XXLUnlabeled data (hrs)=60k, LM Fusion=Yes2020.101.3——
Gen3 Conformer XXL+Unlabeled data (hrs)=60k, LM Fusion=No2020.101.3——
Gen3 Conformer XXL+Unlabeled data (hrs)=60k, LM Fusion=Yes2020.101.3——
Pre-trained Conformer XXLUnlabeled data (hrs)=60k, LM Fusion=Yes2020.101.5——
Gen4 ContextNetUnlabeled data (hrs)=60k, LM Fusion=No2020.101.6——
Gen4 ContextNetUnlabeled data (hrs)=60k, LM Fusion=Yes2020.101.6——
Gen4 Conformer LUnlabeled data (hrs)=60k, LM Fusion=No2020.101.6——
Gen4 Conformer LUnlabeled data (hrs)=60k, LM Fusion=Yes2020.101.6——
Pre-trained CTCUnlabeled data (hrs)=60k, LM Fusion=Yes2020.101.6——
Pre-trained Conformer XLUnlabeled data (hrs)=60k, LM Fusion=Yes2020.101.6——
Pre-trained Conformer XXLUnlabeled data (hrs)=60k, LM Fusion=No2020.101.6——
Pre-trained Conformer XLUnlabeled data (hrs)=60k, LM Fusion=No2020.101.7——
Conformer LUnlabeled data (hrs)=None, LM Fusion=No2020.101.9——
Pre-trained CTCUnlabeled data (hrs)=60k, LM Fusion=No2020.102.1——
ASR + LMSystem=ASR + LM2025.123.35——
ASR + DLMSystem=ASR + DLM2025.123.4——
GLAD-SOTParm.(M)=35.182026.073.5——
SOT-SACTCParm.(M)=35.772026.073.6——
H-SAGEParm.(M)=35.752026.073.6——
SOT + Local MoLEParm.(M)=35.092026.073.7——
ASR onlySystem=ASR only2025.124.08——
SOTParm.(M)=36.072026.074.1——
B2T connectionLayer configuration=Enc-Dec: 6L-6L2022.06—3.698.97
B2T connectionLayer configuration=Enc-Dec: 12L-6L2022.06—3.267.74
Centaurus Base (full SSM)Protocol=Online (streaming), no attention, Parameters (M)=12.4, FLOPs (G)=20.62025.01—5.913.1
Centaurus with causal conv-blockProtocol=Online (streaming), no attention, Parameters (M)=23.6, FLOPs (G)=43.72025.01—4.810.5
Centaurus with FFNProtocol=Online (streaming), no attention, Parameters (M)=18, FLOPs (G)=32.12025.01—5.211.3
Centaurus with Mamba macro-blockProtocol=Online (streaming), no attention, Parameters (M)=29.9, FLOPs (G)=46.92025.01—4.39.9
Full ConvProtocol=Offline (full-context)2025.01—3.19.9
Post-LNLayer configuration=Enc-Dec: 6L-6L2022.06—3.788.76
Pre-LNLayer configuration=Enc-Dec: 6L-6L2022.06—3.899.69
Pre-LNLayer configuration=Enc-Dec: 12L-6L2022.06—3.217.91
Wav2Vec2Protocol=Offline (full-context), Parameters (M)=94.4, FLOPs (G)=2852025.01—1.84.7