TasksSotADatasetsPapersMethodsSubmitAbout
Papers With Code 2

A community resource for machine learning research: papers, code, benchmarks, and state-of-the-art results.

Explore

Notable BenchmarksAll SotADatasetsPapersMethods

Community

Submit ResultsAbout

Data sourced from the PWC Archive (CC-BY-SA 4.0). Built by the community, for the community.

SotA/Audio/Speech Recognition/Switchboard + Hub500

Speech Recognition on Switchboard + Hub500

Metric: Percentage error (lower is better)

LeaderboardDataset
Loading chart...

Results

Submit a result
#Model↕Percentage error▲Extra DataPaperDate↕Code
1IBM (LSTM+Conformer encoder-decoder)4.3NoOn the limit of English conversational speech re...2021-05-03-
2IBM (LSTM encoder-decoder)4.7NoSingle headed attention based sequence-to-sequen...2020-01-20-
3ResNet + BiLSTMs acoustic model5.5NoEnglish Conversational Telephone Speech Recognit...2017-03-06-
4Microsoft 2016b5.8NoAchieving Human Parity in Conversational Speech ...2016-10-17-
5Microsoft 20166.2NoThe Microsoft 2016 Conversational Speech Recogni...2016-09-12-
6VGG/Resnet/LACE/BiLSTM acoustic model trained on SWB+Fisher+CH, N-gram + RNNLM language model trained on Switchboard+Fisher+Gigaword+Broadcast6.3NoThe Microsoft 2016 Conversational Speech Recogni...2016-09-12-
7RNN + VGG + LSTM acoustic model trained on SWB+Fisher+CH, N-gram + "model M" + NNLM language model6.6NoThe IBM 2016 English Conversational Telephone Sp...2016-04-27-
8CNN-LSTM6.6NoAchieving Human Parity in Conversational Speech ...2016-10-17-
9IBM 20166.9NoThe IBM 2016 English Conversational Telephone Sp...2016-04-27-
10RNNLM6.9NoThe Microsoft 2016 Conversational Speech Recogni...2016-09-12-
11IBM 20158NoThe IBM 2015 English Conversational Telephone Sp...2015-05-21-
12HMM-BLSTM trained with MMI + data augmentation (speed) + iVectors + 3 regularizations + Fisher8.5No---
13HMM-TDNN trained with MMI + data augmentation (speed) + iVectors + 3 regularizations + Fisher (10% / 15.1% respectively trained on SWBD only)9.2No---
14CNN on MFSC/fbanks + 1 non-conv layer for FMLLR/I-Vectors concatenated in a DNN10.4No---
15HMM-TDNN + iVectors11No---
16CNN11.5No---
17Deep CNN (10 conv, 4 FC layers), multi-scale feature maps12.2NoVery Deep Multilingual Convolutional Neural Netw...2015-09-29-
18HMM-DNN +sMBR12.6No---
19DNN sMBR12.6No---
20Deep Speech + FSH12.6NoDeep Speech: Scaling up end-to-end speech recogn...2014-12-17Code
21CNN + Bi-RNN + CTC (speech to letters), 25.9% WER if trainedonlyon SWB12.6NoDeep Speech: Scaling up end-to-end speech recogn...2014-12-17Code
22DNN MMI12.9No---
23DNN MPE12.9No---
24DNN BMMI12.9No---
25HMM-TDNN + pNorm + speed up/down speech12.9No---
26DNN + Dropout15NoBuilding DNN Acoustic Models for Large Vocabular...2014-06-30Code
27DNN16NoBuilding DNN Acoustic Models for Large Vocabular...2014-06-30Code
28CD-DNN16.1No---
29DNN-HMM18.5No---
30Deep Speech20NoDeep Speech: Scaling up end-to-end speech recogn...2014-12-17Code