task_path
stringlengths
3
199
dataset
stringlengths
1
128
model_name
stringlengths
1
223
paper_url
stringlengths
21
601
metric_name
stringlengths
1
50
metric_value
stringlengths
1
9.22k
Speech Recognition
LibriSpeech test-clean
Zipformer+pruned transducer w/ CR-CTC (no external language model)
https://arxiv.org/abs/2410.05101v4
Word Error Rate (WER)
1.88
Speech Recognition
LibriSpeech test-clean
ContextNet(L)
https://arxiv.org/abs/2005.03191v3
Word Error Rate (WER)
1.9
Speech Recognition
LibriSpeech test-clean
Conformer(L)
https://arxiv.org/abs/2005.08100v1
Word Error Rate (WER)
1.9
Speech Recognition
LibriSpeech test-clean
Transformer+Time reduction+Self Knowledge distillation
https://arxiv.org/abs/2103.09903v1
Word Error Rate (WER)
1.9
Speech Recognition
LibriSpeech test-clean
ContextNet(M)
https://arxiv.org/abs/2005.03191v3
Word Error Rate (WER)
2
Speech Recognition
LibriSpeech test-clean
Transformer Transducer
https://arxiv.org/abs/2011.03109v2
Word Error Rate (WER)
2.0
Speech Recognition
LibriSpeech test-clean
Conformer(M)
https://arxiv.org/abs/2005.08100v1
Word Error Rate (WER)
2
Speech Recognition
LibriSpeech test-clean
SpeechStew (100M)
https://arxiv.org/abs/2104.02133v3
Word Error Rate (WER)
2.0
Speech Recognition
LibriSpeech test-clean
Qwen-Audio
https://arxiv.org/abs/2311.07919v2
Word Error Rate (WER)
2.0
Speech Recognition
LibriSpeech test-clean
Zipformer+pruned transducer (no external language model)
https://arxiv.org/abs/2310.11230v4
Word Error Rate (WER)
2.00
Speech Recognition
LibriSpeech test-clean
Zipformer+CR-CTC (no external language model)
https://arxiv.org/abs/2410.05101v4
Word Error Rate (WER)
2.02
Speech Recognition
LibriSpeech test-clean
Conv + Transformer AM + Pseudo-Labeling (ConvLM with Transformer Rescoring)
https://arxiv.org/abs/1911.08460v3
Word Error Rate (WER)
2.03
Speech Recognition
LibriSpeech test-clean
Conv + Transformer AM + Iterative Pseudo-Labeling (n-gram LM + Transformer Rescoring)
https://arxiv.org/abs/2005.09267v2
Word Error Rate (WER)
2.10
Speech Recognition
LibriSpeech test-clean
CTC + Transformer LM rescoring
https://arxiv.org/abs/2005.09150v2
Word Error Rate (WER)
2.10
Speech Recognition
LibriSpeech test-clean
Conformer(S)
https://arxiv.org/abs/2005.08100v1
Word Error Rate (WER)
2.1
Speech Recognition
LibriSpeech test-clean
Branchformer + GFSA
https://arxiv.org/abs/2312.04234v5
Word Error Rate (WER)
2.11
Speech Recognition
LibriSpeech test-clean
Multi-Stream Self-Attention With Dilated 1D Convolutions
https://arxiv.org/abs/1910.00716v1
Word Error Rate (WER)
2.20
Speech Recognition
LibriSpeech test-clean
LSTM Transducer
https://arxiv.org/abs/2104.03006v2
Word Error Rate (WER)
2.23
Speech Recognition
LibriSpeech test-clean
Hybrid + Transformer LM rescoring
https://arxiv.org/abs/1910.09799v2
Word Error Rate (WER)
2.26
Speech Recognition
LibriSpeech test-clean
Hybrid model with Transformer rescoring
https://arxiv.org/abs/1905.03072v3
Word Error Rate (WER)
2.3
Speech Recognition
LibriSpeech test-clean
ContextNet(S)
https://arxiv.org/abs/2005.03191v3
Word Error Rate (WER)
2.3
Speech Recognition
LibriSpeech test-clean
Conv + Transformer AM (ConvLM with Transformer Rescoring) (LS only)
https://arxiv.org/abs/1911.08460v3
Word Error Rate (WER)
2.31
Speech Recognition
LibriSpeech test-clean
Squeezeformer (L)
https://arxiv.org/abs/2206.00888v2
Word Error Rate (WER)
2.47
Speech Recognition
LibriSpeech test-clean
LAS + SpecAugment
https://arxiv.org/abs/1904.08779v3
Word Error Rate (WER)
2.5
Speech Recognition
LibriSpeech test-clean
Transformer
https://arxiv.org/abs/1909.06317v2
Word Error Rate (WER)
2.6
Speech Recognition
LibriSpeech test-clean
QuartzNet15x5
https://arxiv.org/abs/1910.10261v1
Word Error Rate (WER)
2.69
Speech Recognition
LibriSpeech test-clean
LAS (no LM)
https://arxiv.org/abs/1904.08779v3
Word Error Rate (WER)
2.7
Speech Recognition
LibriSpeech test-clean
wav2vec_wav2letter
https://arxiv.org/abs/2010.11430v1
Word Error Rate (WER)
2.7
Speech Recognition
LibriSpeech test-clean
Espresso
https://arxiv.org/abs/1909.08723v3
Word Error Rate (WER)
2.8
Speech Recognition
LibriSpeech test-clean
Jasper DR 10x5 (+ Time/Freq Masks)
https://arxiv.org/abs/1904.03288v3
Word Error Rate (WER)
2.84
Speech Recognition
LibriSpeech test-clean
Jasper DR 10x5
https://arxiv.org/abs/1904.03288v3
Word Error Rate (WER)
2.95
Speech Recognition
LibriSpeech test-clean
tdnn + chain + rnnlm rescoring
https://www.cs.jhu.edu/~hxu/neural-network-language.pdf
Word Error Rate (WER)
3.06
Speech Recognition
LibriSpeech test-clean
Convolutional Speech Recognition
http://arxiv.org/abs/1812.06864v2
Word Error Rate (WER)
3.26
Speech Recognition
LibriSpeech test-clean
MT4SSL
https://arxiv.org/abs/2211.07321v3
Word Error Rate (WER)
3.4
Speech Recognition
LibriSpeech test-clean
Model Unit Exploration
https://arxiv.org/abs/1902.01955v2
Word Error Rate (WER)
3.60
Speech Recognition
LibriSpeech test-clean
Seq-to-seq attention
http://arxiv.org/abs/1805.03294v1
Word Error Rate (WER)
3.82
Speech Recognition
LibriSpeech test-clean
CTC-CRF 4gram-LM
https://ieeexplore.ieee.org/document/8682256
Word Error Rate (WER)
4.09
Speech Recognition
LibriSpeech test-clean
HMM-TDNN trained with MMI + data augmentation (speed) + iVectors + 3 regularizations
null
Word Error Rate (WER)
4.3
Speech Recognition
LibriSpeech test-clean
Centaurus (30 M)
https://arxiv.org/abs/2501.13230v1
Word Error Rate (WER)
4.4
Speech Recognition
LibriSpeech test-clean
HMM-TDNN + iVectors
null
Word Error Rate (WER)
4.8
Speech Recognition
LibriSpeech test-clean
Gated ConvNets
http://arxiv.org/abs/1712.09444v2
Word Error Rate (WER)
4.8
Speech Recognition
LibriSpeech test-clean
Deep Speech 2
http://arxiv.org/abs/1512.02595v1
Word Error Rate (WER)
5.33
Speech Recognition
LibriSpeech test-clean
CTC + policy learning
http://arxiv.org/abs/1712.07101v1
Word Error Rate (WER)
5.42
Speech Recognition
LibriSpeech test-clean
HMM-DNN + pNorm*
null
Word Error Rate (WER)
5.5
Speech Recognition
LibriSpeech test-clean
Li-GRU
http://arxiv.org/abs/1811.07453v2
Word Error Rate (WER)
6.2
Speech Recognition
LibriSpeech test-clean
Snips
http://arxiv.org/abs/1805.10190v3
Word Error Rate (WER)
6.4
Speech Recognition
LibriSpeech test-clean
Local Prior Matching (Large Model)
https://arxiv.org/abs/2002.10336v1
Word Error Rate (WER)
7.19
Speech Recognition
LibriSpeech test-clean
HMM-(SAT)GMM
null
Word Error Rate (WER)
8.0
Speech Recognition
LibriSpeech test-clean
AmNet
https://arxiv.org/abs/2108.01553v1
Word Error Rate (WER)
8.6
Speech Recognition
GigaSpeech TEST
Zipformer+pruned transducer w/ CR-CTC (no external language model)
https://arxiv.org/abs/2410.05101v4
Word Error Rate (WER)
10.03
Speech Recognition
GigaSpeech TEST
Zipformer+CR-CTC/AED (no external language model)
https://arxiv.org/abs/2410.05101v4
Word Error Rate (WER)
10.07
Speech Recognition
GigaSpeech TEST
Zipformer+pruned transducer (no external language model)
https://arxiv.org/abs/2410.05101v4
Word Error Rate (WER)
10.2
Speech Recognition
GigaSpeech TEST
Zipformer+CR-CTC (no external language model)
https://arxiv.org/abs/2410.05101v4
Word Error Rate (WER)
10.28
Speech Recognition
GigaSpeech TEST
Conformer/Transformer-AED
https://arxiv.org/abs/2106.06909v1
Word Error Rate (WER)
10.80
Speech Recognition
VietMed
XLSR-53-Viet
https://arxiv.org/abs/2404.05659v3
Dev WER
26.8
Speech Recognition
VietMed
XLSR-53-Viet
https://arxiv.org/abs/2404.05659v3
Test WER
29.6
Speech Recognition
VietMed
XLSR-53
https://arxiv.org/abs/2404.05659v3
Dev WER
45.2
Speech Recognition
VietMed
XLSR-53
https://arxiv.org/abs/2404.05659v3
Test WER
51.8
Speech Recognition
VietMed
w2v2-Viet
https://arxiv.org/abs/2404.05659v3
Dev WER
45.3
Speech Recognition
VietMed
w2v2-Viet
https://arxiv.org/abs/2404.05659v3
Test WER
49.5
Speech Recognition
VietMed
GMM-HMM SAT+VTLN
https://arxiv.org/abs/2404.05659v3
Dev WER
52.2
Speech Recognition
VietMed
GMM-HMM SAT
https://arxiv.org/abs/2404.05659v3
Dev WER
52.6
Speech Recognition
VietMed
GMM-HMM Tri
https://arxiv.org/abs/2404.05659v3
Dev WER
61.3
Speech Recognition
VietMed
GMM-HMM VTLN
https://arxiv.org/abs/2404.05659v3
Dev WER
61.3
Speech Recognition
VietMed
GMM-HMM Mono
https://arxiv.org/abs/2404.05659v3
Dev WER
71.7
Speech Recognition
Europarl-ASR EN Guest-test
United-MedASR (764M)
https://arxiv.org/abs/2412.00055v1
WER
0.26
Speech Recognition
Europarl-ASR EN Guest-test
mllp_2021_offline_verb
https://www.isca-speech.org/archive/interspeech_2021/diazmunio21_interspeech.html
WER
7.0
Speech Recognition
Europarl-ASR EN Guest-test
mllp_2021_streaming_verb
https://www.isca-speech.org/archive/interspeech_2021/diazmunio21_interspeech.html
WER
7.3
Speech Recognition
Switchboard + Hub500
IBM (LSTM+Conformer encoder-decoder)
https://arxiv.org/abs/2105.00982v1
Percentage error
4.3
Speech Recognition
Switchboard + Hub500
IBM (LSTM encoder-decoder)
https://arxiv.org/abs/2001.07263v3
Percentage error
4.7
Speech Recognition
Switchboard + Hub500
ResNet + BiLSTMs acoustic model
http://arxiv.org/abs/1703.02136v1
Percentage error
5.5
Speech Recognition
Switchboard + Hub500
Microsoft 2016b
http://arxiv.org/abs/1610.05256v2
Percentage error
5.8
Speech Recognition
Switchboard + Hub500
Microsoft 2016
http://arxiv.org/abs/1609.03528v2
Percentage error
6.2
Speech Recognition
Switchboard + Hub500
VGG/Resnet/LACE/BiLSTM acoustic model trained on SWB+Fisher+CH, N-gram + RNNLM language model trained on Switchboard+Fisher+Gigaword+Broadcast
http://arxiv.org/abs/1609.03528v2
Percentage error
6.3
Speech Recognition
Switchboard + Hub500
RNN + VGG + LSTM acoustic model trained on SWB+Fisher+CH, N-gram + "model M" + NNLM language model
http://arxiv.org/abs/1604.08242v2
Percentage error
6.6
Speech Recognition
Switchboard + Hub500
CNN-LSTM
http://arxiv.org/abs/1610.05256v2
Percentage error
6.6
Speech Recognition
Switchboard + Hub500
IBM 2016
http://arxiv.org/abs/1604.08242v2
Percentage error
6.9
Speech Recognition
Switchboard + Hub500
RNNLM
http://arxiv.org/abs/1609.03528v2
Percentage error
6.9
Speech Recognition
Switchboard + Hub500
IBM 2015
http://arxiv.org/abs/1505.05899v1
Percentage error
8.0
Speech Recognition
Switchboard + Hub500
HMM-BLSTM trained with MMI + data augmentation (speed) + iVectors + 3 regularizations + Fisher
null
Percentage error
8.5
Speech Recognition
Switchboard + Hub500
HMM-TDNN trained with MMI + data augmentation (speed) + iVectors + 3 regularizations + Fisher (10% / 15.1% respectively trained on SWBD only)
null
Percentage error
9.2
Speech Recognition
Switchboard + Hub500
CNN on MFSC/fbanks + 1 non-conv layer for FMLLR/I-Vectors concatenated in a DNN
null
Percentage error
10.4
Speech Recognition
Switchboard + Hub500
HMM-TDNN + iVectors
null
Percentage error
11
Speech Recognition
Switchboard + Hub500
CNN
null
Percentage error
11.5
Speech Recognition
Switchboard + Hub500
Deep CNN (10 conv, 4 FC layers), multi-scale feature maps
http://arxiv.org/abs/1509.08967v2
Percentage error
12.2
Speech Recognition
Switchboard + Hub500
HMM-DNN +sMBR
null
Percentage error
12.6
Speech Recognition
Switchboard + Hub500
DNN sMBR
null
Percentage error
12.6
Speech Recognition
Switchboard + Hub500
Deep Speech + FSH
http://arxiv.org/abs/1412.5567v2
Percentage error
12.6
Speech Recognition
Switchboard + Hub500
CNN + Bi-RNN + CTC (speech to letters), 25.9% WER if trainedonlyon SWB
http://arxiv.org/abs/1412.5567v2
Percentage error
12.6
Speech Recognition
Switchboard + Hub500
DNN MMI
null
Percentage error
12.9
Speech Recognition
Switchboard + Hub500
DNN MPE
null
Percentage error
12.9
Speech Recognition
Switchboard + Hub500
DNN BMMI
null
Percentage error
12.9
Speech Recognition
Switchboard + Hub500
HMM-TDNN + pNorm + speed up/down speech
null
Percentage error
12.9
Speech Recognition
Switchboard + Hub500
DNN + Dropout
http://arxiv.org/abs/1406.7806v2
Percentage error
15
Speech Recognition
Switchboard + Hub500
DNN
http://arxiv.org/abs/1406.7806v2
Percentage error
16
Speech Recognition
Switchboard + Hub500
CD-DNN
null
Percentage error
16.1
Speech Recognition
Switchboard + Hub500
DNN-HMM
null
Percentage error
18.5
Speech Recognition
Switchboard + Hub500
Deep Speech
http://arxiv.org/abs/1412.5567v2
Percentage error
20
Speech Recognition
LibriSpeech train-clean-100 test-clean
wav2vec_wav2letter
https://arxiv.org/abs/2010.11430v1
Word Error Rate (WER)
2.8
Speech Recognition
LibriSpeech test-other
SAMBA ASR
https://arxiv.org/abs/2501.02832v3
Word Error Rate (WER)
2.48