VGG/Resnet/LACE/BiLSTM acoustic model trained on SWB+Fisher+CH, N-gram + RNNLM language model trained on Switchboard+Fisher+Gigaword+Broadcast

Reported on 2 benchmarks across 1 task · 1 paper · 2 SOTA

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

Audio2 results

Speech Recognitiononswb_hub_500 WER fullSWBCH
Percentage error· 2016-09-12
11.9
best: 6.8 (IBM (LSTM+Conformer encoder-decoder))
SOTA
The Microsoft 2016 Conversational Speech Recognition System arXiv:1609.03528
Speech RecognitiononSwitchboard + Hub500
Percentage error· 2016-09-12
6.3
best: 4.3 (IBM (LSTM+Conformer encoder-decoder))
SOTA
The Microsoft 2016 Conversational Speech Recognition System arXiv:1609.03528