Browse › Audio › Speech Recognition › LibriSpeech test-other
LibriSpeech test-other Benchmark (Speech Recognition)
Speech Recognition is the task of converting spoken language into text. It involves recognizing the words spoken in an audio recording and transcribing them into a written format. The goal is to accurately transcribe the speech in real-time or from recorded audio, taking into account factors such as accents, speaking speed, and background noise.
The archive carries no text for this table; the description above is the archive's text for the task Speech Recognition. archive 2025-07-28
Over time archive 2025-07-28
The chart needs JavaScript; the table below carries every value.
Direction inferred from the metric name, not from the archive: Word Error Rate (WER) (lower is better). Points are placed at the row's paper date; 52 of 53 rows carry one.
Results archive 2025-07-28
Archive rows end at the archive snapshot, 2025-07-28: no result published after that date is in this table. Rank is the archive's row order at that snapshot; not re-ranked here. Metric values are the archive's strings. Column headers sort the table in your browser; each row keeps its archive rank.
| Paper | Code | Ran Syntology | Report | |||||
|---|---|---|---|---|---|---|---|---|
| 1 | SAMBA ASR | 2.48 | – | Paper | – | 2025 | no code linked | report |
| 2 | FAdam | 2.49 | – | Paper | Code | 2024 | linked, not harvested | report |
| 3 | w2v-BERT XXL | 2.5 | – | Paper | Code | 2021 | linked, not harvested | report |
| 4 | Conformer + Wav2vec 2.0 + SpecAugment-based Noisy Student Training with Libri-Light | 2.6 | – | Paper | Code | 2020 | 2 of 2 ran · 0 unverified | report |
| 5 | HuBERT with Libri-Light | 2.9 | – | Paper | Code | 2021 | 0 of 9 ran · 9 unverified | report |
| 6 | wav2vec 2.0 with Libri-Light | 3.0 | – | Paper | Code | 2020 | 2 of 9 ran · 7 unverified | report |
| 7 | Conv + Transformer + wav2vec2.0 + pseudo labeling | 3.1 | – | Paper | Code | 2020 | linked, not harvested | report |
| 8 | WavLM Large | 3.2 | – | Paper | Code | 2021 | linked, not harvested | report |
| 9 | SpeechStew (1B) | 3.3 | – | Paper | – | 2021 | no code linked | report |
| 10 | ContextNet + SpecAugment-based Noisy Student Training with Libri-Light | 3.4 | – | Paper | Code | 2020 | linked, not harvested | report |
| 11 | E-Branchformer (L) + Internal Language Model Estimation | 3.65 | – | Paper | Code | 2022 | linked, not harvested | report |
| 12 | data2vec | 3.7 | – | Paper | Code | 2022 | 0 of 6 ran · 6 unverified | report |
| 13 | Conv + Transformer AM + Iterative Pseudo-Labeling (n-gram LM + Transformer Rescoring) | 3.83 | – | Paper | Code | 2020 | linked, not harvested | report |
| 14 | Conformer(L) | 3.9 | ✓ | Paper | Code | 2020 | 4 of 7 ran · 3 unverified | report |
| 15 | Zipformer+pruned transducer w/ CR-CTC (no external language model) | 3.95 | – | Paper | Code | 2024 | linked, not harvested | report |
| 16 | SpeechStew (100M) | 4.0 | – | Paper | – | 2021 | no code linked | report |
| 17 | wav2vec 2.0 | 4.1 | ✓ | Paper | Code | 2020 | 2 of 9 ran · 7 unverified | report |
| 18 | ContextNet(L) | 4.1 | – | Paper | Code | 2020 | linked, not harvested | report |
| 19 | Conv + Transformer AM (ConvLM with Transformer Rescoring) | 4.11 | ✓ | Paper | Code | 2019 | linked, not harvested | report |
| 20 | CTC + Transformer LM rescoring | 4.20 | ✓ | Paper | – | 2020 | no code linked | report |
| 21 | Transformer Transducer | 4.20 | ✓ | Paper | Code | 2020 | linked, not harvested | report |
| 22 | Qwen-Audio | 4.2 | – | Paper | Code | 2023 | 5 of 7 ran · 2 unverified | report |
| 23 | Conformer(M) | 4.3 | ✓ | Paper | Code | 2020 | 4 of 7 ran · 3 unverified | report |
| 24 | Zipformer+CR-CTC (no external language model) | 4.35 | – | Paper | Code | 2024 | linked, not harvested | report |
| 25 | Zipformer+pruned transducer (no external language model) | 4.38 | – | Paper | Code | 2023 | linked, not harvested | report |
| 26 | Multistream CNN with Self-Attentive SRU | 4.46 | – | Paper | – | 2020 | no code linked | report |
| 27 | ContextNet(M) | 4.5 | ✓ | Paper | Code | 2020 | linked, not harvested | report |
| 28 | hybrid + Transformer LM rescoring | 4.85 | ✓ | Paper | – | 2019 | no code linked | report |
| 29 | Branchformer + GFSA | 4.94 | – | Paper | Code | 2023 | 19 of 29 ran · 10 unverified | report |
| 30 | Hybrid model with Transformer rescoring | 5.0 | – | Paper | Code | 2019 | linked, not harvested | report |
| 31 | Conformer(S) | 5.0 | ✓ | Paper | Code | 2020 | 4 of 7 ran · 3 unverified | report |
| 32 | Conv + Transformer AM (ConvLM with Transformer Rescoring) (LS only) | 5.18 | – | Paper | Code | 2019 | linked, not harvested | report |
| 33 | ContextNet(S) | 5.5 | ✓ | Paper | Code | 2020 | linked, not harvested | report |
| 34 | LSTM Transducer | 5.6 | ✓ | Paper | Code | 2021 | linked, not harvested | report |
| 35 | Transformer | 5.7 | ✓ | Paper | Code | 2019 | linked, not harvested | report |
| 36 | LAS + SpecAugment | 5.8 | ✓ | Paper | Code | 2019 | 1 of 18 ran · 17 unverified | report |
| 37 | Multi-Stream Self-Attention With Dilated 1D Convolutions | 5.80 | – | Paper | Code | 2019 | 1 of 1 ran · 0 unverified | report |
| 38 | Squeezeformer (L) | 5.97 | – | Paper | Code | 2022 | 31 of 49 ran · 18 unverified | report |
| 39 | LAS (no LM) | 6.5 | ✓ | Paper | Code | 2019 | 1 of 18 ran · 17 unverified | report |
| 40 | Conformer with Relaxed Attention | 6.85 | – | Paper | Code | 2021 | linked, not harvested | report |
| 41 | QuartzNet15x5 | 7.25 | – | Paper | Code | 2019 | 3 of 12 ran · 9 unverified | report |
| 42 | tdnn + chain + rnnlm rescoring | 7.63 | ✓ | Paper | – | 2018 | no code linked | report |
| 43 | Jasper DR 10x5 (+ Time/Freq Masks) | 7.84 | – | Paper | Code | 2019 | linked, not harvested | report |
| 44 | Espresso | 8.7 | – | Paper | Code | 2019 | 1 of 1 ran · 0 unverified | report |
| 45 | Jasper DR 10x5 | 8.79 | – | Paper | Code | 2019 | linked, not harvested | report |
| 46 | MT4SSL | 9.6 | – | Paper | Code | 2022 | linked, not harvested | report |
| 47 | Convolutional Speech Recognition | 10.47 | ✓ | Paper | – | 2018 | no code linked | report |
| 48 | CTC-CRF 4gram-LM | 10.65 | – | Paper | Code | 2019 | linked, not harvested | report |
| 49 | TDNN + pNorm + speed up/down speech | 12.5 | – | – | – | not matched | report | |
| 50 | Deep Speech 2 | 13.25 | – | Paper | Code | 2015 | 2 of 39 ran · 37 unverified | report |
| 51 | Local Prior Matching (Large Model, ConvLM LM) | 15.28 | – | Paper | Code | 2020 | linked, not harvested | report |
| 52 | Snips | 16.5 | – | Paper | Code | 2018 | 1 of 15 ran · 14 unverified | report |
| 53 | Local Prior Matching (Large Model) | 20.84 | ✓ | Paper | Code | 2020 | linked, not harvested | report |
All 53 rows shown. 52 link to a paper page on this site; 17 are marked as using additional training data in the archive. No GitHub stars are tracked; "Code" is the first repository the archive lists for the row. The archive carries no row tags, review links or community-submitted rows for this table; none are shown. archive 2025-07-28
Syntology Ran reads "N of M ran · U unverified": of the M code samples Syntology harvested from repositories linked to that row's paper (joined by arXiv id), N executed on a synthesized input and the other U = M−N are unverified (harvested, no recorded run). It counts code from repositories linked to that row's paper, not this result: the row's number was not reproduced and nothing here is a correctness claim. The other cell texts mean no graph line for the row: "linked, not harvested" (the archive links code, Syntology has not harvested it), "no code linked" (no code link in the archive), "not matched" (the row's paper URL matched no paper on this site). 18 rows have a graph line, from 14 distinct papers; 16 rows (12 papers) have at least one sample that ran. Counting each paper once: Syntology ran 72 of 204 samples; 132 unverified. Separately, 17 of those 204 are pointer-only (licence): the site points at that code rather than redistributing it, a licence property recorded for ran and unverified samples alike; each cell's tooltip carries the row's own pointer-only count. Read from the graph 2026-09-24. Per-sample status is on the paper page.
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections