Codesota · Benchmark · LibriSpeechHome/Leaderboards/Audio & Speech/Automatic Speech Recognition/LibriSpeech
Johns Hopkins University

LibriSpeech.

1000 hours of English speech from audiobooks. Standard benchmark for automatic speech recognition with clean and noisy test splits.

Paper ↗Leaderboard ↓Lineage
§ 01 · SOTA history

Year over year.

§ 02 · Leaderboard

Results by metric.

Found a wrong score or missing run?
Use row edits to send a sourced correction into moderation.
Add / edit result ↗Report issue ↗

Wer

Wer is the reported evaluation metric for LibriSpeech. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.

Lower is better

Trust tiers for Werverifiedpapervendorcommunityunverified

Muted rows were not state of the art when published — an earlier or same-year result already scored better.

RankModelTrustScoreYearLinksFix
01Granite 4.0 1B Speechunverified2.852025Paper ↗Source ↗Looks wrong?
02Granite Speech 3.3 8Bunverified2.862025Paper ↗Source ↗Looks wrong?
03Canary-1B-Flashunverified2.872025Paper ↗Source ↗Looks wrong?
04Canary-1Bunverified2.932024Paper ↗Looks wrong?
05Distil-Whisper Large v2unverified2.942023Paper ↗Code ↗Looks wrong?
06Whisper Medium (English)unverified3.022022Paper ↗Code ↗Looks wrong?
07Whisper-small.enunverified3.052022Paper ↗Code ↗Source ↗Looks wrong?
08Whisper Small (English)unverified3.052022Paper ↗Code ↗Looks wrong?
09Llama 3 Speech (70B)unverified3.102024Paper ↗Code ↗Looks wrong?
10Llama 3 (405B, Instruct)unverified3.102024Paper ↗Code ↗Looks wrong?
11Canary-Qwen-2.5Bunverified3.102025Paper ↗Source ↗Looks wrong?
12Parakeet-tdt-0.6b-v2unverified3.192023Paper ↗Code ↗Source ↗Looks wrong?
13Granite Speech 3.3 2Bunverified3.262025Paper ↗Source ↗Looks wrong?
14Voxtral-Small-24B-2507unverified3.262025Paper ↗Source ↗Looks wrong?
15Moonshine-baseunverified3.382024Paper ↗Code ↗Source ↗Looks wrong?
16Distil-Whisper Small (English)unverified3.482023Paper ↗Code ↗Looks wrong?
17Parakeet-ctc-1.1bunverified3.512023Paper ↗Source ↗Looks wrong?
18Canary-1b-v2unverified3.562025Paper ↗Looks wrong?
19Parakeet-tdt-0.6b-v3unverified3.592023Paper ↗Code ↗Source ↗Looks wrong?
20Distil-Whisper Medium (English)unverified3.692023Paper ↗Code ↗Looks wrong?
21Parakeet-ctc-0.6bunverified3.802023Paper ↗Source ↗Looks wrong?
22Phi-4 Multimodal Instructunverified3.822025Paper ↗Source ↗Looks wrong?
23Asr-wav2vec2-librispeechunverified3.832021Paper ↗Code ↗Source ↗Looks wrong?
24Lite-whisper-large-v3-accunverified3.912025Paper ↗Code ↗Source ↗Looks wrong?
25Whisper Large v3unverified3.912022Paper ↗Code ↗Source ↗Looks wrong?
26Stt_en_fastconformer_transducer_largeunverified3.972023Paper ↗Source ↗Looks wrong?
27Stt_en_fastconformer_ctc_largeunverified4.042023Paper ↗Source ↗Looks wrong?
28Stt_en_conformer_ctc_largeunverified4.152020Paper ↗Code ↗Source ↗Looks wrong?
29Whisper Large v3 Turbounverified4.242022Paper ↗Code ↗Source ↗Looks wrong?
30Asr-conformer-loquaciousunverified4.242025Paper ↗Looks wrong?
31Whisper baseunverified4.252022Paper ↗Code ↗Looks wrong?
32Canary-180M-Flashunverified4.352025Paper ↗Source ↗Looks wrong?
33Lite-whisper-large-v3unverified4.402025Paper ↗Code ↗Source ↗Looks wrong?
34Qwen3-ASR-0.6Bunverified4.452026Paper ↗Code ↗Source ↗Looks wrong?
35SYMPHONYunverified4.482025Paper ↗Looks wrong?
36Moonshine-streaming-tinyunverified4.502026Paper ↗Looks wrong?
37Moonshine-tinyunverified4.552024Paper ↗Code ↗Source ↗Looks wrong?
38Lite-whisper-large-v3-turbo-accunverified4.602025Paper ↗Code ↗Source ↗Looks wrong?
39Owsm_ctc_v4_1Bunverified4.892025Paper ↗Code ↗Source ↗Looks wrong?
40Moonshine Streaming Mediumunverified5.002026Paper ↗Source ↗Looks wrong?
41Distil-large-v3.5unverified5.042023Paper ↗Code ↗Source ↗Looks wrong?
42Zipformer-transducer-XL-290Munverified5.042023Paper ↗Code ↗Source ↗Looks wrong?
43Whisper Large v2unverified5.142022Paper ↗Code ↗Source ↗Looks wrong?
44Owsm_ctc_v3.1_1Bunverified5.152024Paper ↗Code ↗Source ↗Looks wrong?
45Distil-large-v3unverified5.192023Paper ↗Code ↗Source ↗Looks wrong?
46Lite-whisper-large-v3-fastunverified5.192025Paper ↗Code ↗Source ↗Looks wrong?
47Parakeet-tdt_ctc-110munverified5.222023Paper ↗Code ↗Source ↗Looks wrong?
48Voxtral-Mini-4B-Realtime-2602unverified5.522026Paper ↗Source ↗Looks wrong?
49Whisper Largeunverified5.542022Paper ↗Code ↗Source ↗Looks wrong?
50Whisper Tiny (English)unverified5.662022Paper ↗Code ↗Looks wrong?
51Moshi ASRunverified5.702024Paper ↗Code ↗Looks wrong?
52Whisper-medium.enunverified5.852022Paper ↗Code ↗Source ↗Looks wrong?
53Moonshine-streaming-smallunverified6.782026Paper ↗Source ↗Looks wrong?
54Distil-large-v2unverified6.842023Paper ↗Code ↗Source ↗Looks wrong?
55Distil-small.enunverified7.732023Paper ↗Code ↗Source ↗Looks wrong?
56Stt_en_conformer_ctc_smallunverified7.922020Paper ↗Code ↗Source ↗Looks wrong?
57Distil-medium.enunverified8.352023Paper ↗Code ↗Source ↗Looks wrong?
58Niagara-38m-batch.enunverified9.352026Paper ↗Source ↗Looks wrong?
59Whisper-base.enunverified10.352022Paper ↗Code ↗Source ↗Looks wrong?
60Niagara-19m-batch.enunverified11.22026Paper ↗Looks wrong?
61Hubert-xlarge-ls960-ftunverified12.222021Paper ↗Code ↗Source ↗Looks wrong?
62Wav2vec2-large-960h-lv60-selfunverified12.422020Paper ↗Code ↗Source ↗Looks wrong?
63Wav2vec2-conformer-rel-pos-large-960h-ftunverified12.442020Paper ↗Code ↗Source ↗Looks wrong?
64Wav2vec2-base-960hunverified12.532020Paper ↗Code ↗Source ↗Looks wrong?
65Wav2vec2-conformer-rope-large-960h-ftunverified12.542020Paper ↗Code ↗Source ↗Looks wrong?
66Mms-1b-allunverified12.632023Paper ↗Code ↗Source ↗Looks wrong?
67Hubert-large-ls960-ftunverified12.752021Paper ↗Code ↗Source ↗Looks wrong?
68Data2vec-audio-large-960hunverified12.942022Paper ↗Code ↗Source ↗Looks wrong?
69Wav2vec2-large-robust-ft-libri-960hunverified13.762021Paper ↗Code ↗Source ↗Looks wrong?
70Whisper-tiny.enunverified15.452022Paper ↗Code ↗Source ↗Looks wrong?
71wav2vec 2.0 Large (960h)unverified15.462020Paper ↗Code ↗Source ↗Looks wrong?
72Data2vec-audio-base-960hunverified15.482022Paper ↗Code ↗Source ↗Looks wrong?
73Mms-1b-fl102unverified28.72023Paper ↗Code ↗Source ↗Looks wrong?

WER (test-other)

Word Error Rate on noisier/accented speech (lower is better)

Lower is better

Trust tiers for WER (test-other)verifiedpapervendorcommunityunverified

Muted rows were not state of the art when published — an earlier or same-year result already scored better.

RankModelTrustScoreYearLinksFix
01Universal-1
Greedy decoding, no external LM. From AssemblyAI research page Table 1.
verified3.102024Paper ↗Looks wrong?
02Parakeet TDT 0.6B v2
NVIDIA. 0.6B params. FastConformer-TDT.
paper3.192025Source ↗Looks wrong?
03wav2vec 2.0 Large (960h)
test-other WER (%). wav2vec 2.0 Large, 960h. Source: Table 3, arxiv:2006.11477
verified3.302020Paper ↗Looks wrong?
04Canary 1B v2
NVIDIA. 1B multilingual ASR+AST. Aug 2025.
paper3.562025Source ↗Looks wrong?
05Parakeet TDT 0.6B v3
NVIDIA. 0.6B params. Multilingual. Sep 2025.
paper3.592025Source ↗Looks wrong?
06Whisper Large v3
test-other WER (%). Whisper large-v3. Source: OpenAI model card / arxiv:2212.04356
paper3.602024Source ↗Looks wrong?
07HuBERT Large (LS-960)
test-other WER (%). HuBERT Large, 960h. Source: Table 2, arxiv:2106.07447
verified3.602021Paper ↗Looks wrong?
08Canary-1B
test-other WER (%). Canary-1B EN. Source: Table 2, arxiv:2310.09873
verified3.802026Source ↗Looks wrong?
09Voxtral Mini 3B
Mistral AI. 3B multimodal model. July 2025.
paper4.082025Source ↗Looks wrong?
10Google USM
test-other WER (%). Google USM 2B. Source: Table 3, arxiv:2303.01037
verified4.102023Paper ↗Looks wrong?
11Parakeet-CTC-1.1B
test-other WER (%). Parakeet-CTC-1.1B. Source: Table 1, arxiv:2311.13251
verified4.202026Source ↗Looks wrong?
12Whisper Large v2
test-other WER (%). Whisper large-v2. Source: Table 5, arxiv:2212.04356
verified5.202026Source ↗Looks wrong?
13Pulse STT
Smallest AI docs: English STT - ESB Dataset (Streaming), LibriSpeech Other WER.
verified5.832026Source ↗Looks wrong?
14Phi-4-multimodal-instruct
Microsoft. 5.6B multimodal model. Feb 2025.
paper5.972025Source ↗Looks wrong?

WER (test-clean)

Wer Test Clean is the reported evaluation metric for LibriSpeech. Codesota tracks published model scores on this metric so readers can compare state-of-the-art results across sources and model families.

Lower is better

Trust tiers for WER (test-clean)verifiedpapervendorcommunityunverified

Muted rows were not state of the art when published — an earlier or same-year result already scored better.

RankModelTrustScoreYearLinksFix
01Whisper Large v2
test-clean WER (%). Whisper large-v2. Source: Table 5, arxiv:2212.04356
verified2.702022Paper ↗Looks wrong?
02wav2vec 2.0 Large
Meta. 317M params. Self-supervised pre-training on 60k hours of speech. Foundational SSL ASR model.
paper2.902024Source ↗Looks wrong?
03Pulse STT
Smallest AI docs: English STT - ESB Dataset (Streaming), LibriSpeech Clean WER.
verified3.222026Source ↗Looks wrong?
Lineage

LibriSpeech in context.

See full speech recognition benchmarks lineage →
None — this is where the lineage begins.
This benchmark (1)
saturated2015-04
LibriSpeech
Successors (3)
active2017-06
VoxCeleb
VoxCeleb covers speaker identity, not transcription — a different task that addresses the 'who spoke' question LibriSpeech ignores. Speaker verification became a standard parallel track in speech evaluation.
active2020-04
CHiME-6
LibriSpeech test-other saturated; CHiME-6's multi-speaker dinner-party setup was the first major challenge where clean-speech progress didn't transfer. Where attention moved when LibriSpeech-other WER dropped below 4%.
active2021-06
GigaSpeech
GigaSpeech is a scale and diversity extension — 10× more data, multi-domain. A training and evaluation resource for robustness rather than a direct successor to LibriSpeech's narrow clean-speech task.
§ 04 · Submit a result

Add to the leaderboard.

← Back to Automatic Speech Recognition