List your company

Automatic Speech Recognition on Eval2000-CH Fisher-Switchboard 2300-h (test)

9WER (SW Subset)CTC/attention
8.76810.33411.913.466May 27, 2020

Evaluation results

MethodLinks
CTC/attentionUnit=BPE, LM=RNN2020.059—18.113.6———
LF-MMIUnit=bi-phone, LM=4-gram2020.059.5—18.614.18.317.112.7
E2E-LF-MMIUnit=bi-phone, LM=4-gram2020.059.8—19.314.68.517.413
CATUnit=mono-phone, LM=4-gram2020.059.8—18.814.38.817.413.1
LF-MMIUnit=mono-phone, LM=4-gram2020.0510.7—20.315.5———
LASUnit=subword, LM=RNN2020.0510.9—19.415.2———
E2E-LF-MMIUnit=mono-phone, LM=4-gram2020.0511—20.715.9———
AttentionUnit=subword, LM=RNN2020.0511.8—25.718.1———
AttentionUnit=subword, LM=No LM2020.0513.5—27.120.3———
EESENUnit=mono-phone, LM=3-gram2020.0514.8—2620.4———
AttentionUnit=char, LM=No LM, RNN-LM rescoring=false2020.05—15.5—————
CATUnit=mono-phone, LM=4-gram, RNN-LM rescoring=false2020.05—16—————
CATUnit=mono-phone, LM=4-gram, RNN-LM rescoring=true2020.05—15—————
E2E-LF-MMIUnit=bi-phone, LM=4-gram, RNN-LM rescoring=false2020.05—15.4—————
E2E-LF-MMIUnit=bi-phone, LM=4-gram, RNN-LM rescoring=true2020.05—14.5—————
E2E-LF-MMIUnit=mono-phone, LM=4-gram, RNN-LM rescoring=false2020.05—16.8—————
LF-MMIUnit=bi-phone, LM=4-gram, RNN-LM rescoring=false2020.05—15.1—————
LF-MMIUnit=bi-phone, LM=4-gram, RNN-LM rescoring=true2020.05—14.3—————
RNN-TUnit=char, LM=4-gram, RNN-LM rescoring=false2020.05—17.5—————