Convolutional Speech Recognition

Reported on 5 benchmarks across 1 task · 1 paper · 3 SOTA

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

Audio6 results

Speech RecognitiononWSJ dev93
Word Error Rate (WER)· 2018-12-17
6.8
best: 5.68 (CTC-CRF ST-NAS)
SOTA
Fully Convolutional Speech Recognition arXiv:1812.06864
Speech RecognitiononLibriSpeech test-clean
Word Error Rate (WER)· uses extra data· 2018-12-17
3.26
best: 0.985 (United Med ASR)
SOTA
Fully Convolutional Speech Recognition arXiv:1812.06864
Speech RecognitiononLibriSpeech test-other
Word Error Rate (WER)· uses extra data· 2018-12-17
10.47
best: 2.48 (SAMBA ASR)
SOTA
Fully Convolutional Speech Recognition arXiv:1812.06864
Speech RecognitiononWSJ eval92
Word Error Rate (WER)· 2018-12-17
3.5
best: 1.3 (Speechstew 100M)
Fully Convolutional Speech Recognition arXiv:1812.06864
Speech RecognitiononWSJ eval93
Word Error Rate (WER)· 2018-12-17
6.8
best: 4.98 (Deep Speech 2)
Fully Convolutional Speech Recognition arXiv:1812.06864
Speech RecognitiononWSJ dev93
Word Error Rate (WER)
6.23
best: 5.68 (CTC-CRF ST-NAS)