Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Automatic Speech Recognition on LibriSpeech 100h clean (dev)
Loading...
2.1
WER
SpeechT5
1.624
4.837
8.05
11.263
May 19, 2020
May 22, 2021
May 26, 2022
May 30, 2023
Jun 1, 2024
Jun 5, 2025
Jun 9, 2026
WER
Updated 1mo ago
Evaluation Results
Method
Method
Links
WER
SpeechT5
LM=Transformer
2021.10
2.1
wav2vec 2.0 BASE
LM=Transformer
2021.10
2.2
Baseline
LM=Transformer
2021.10
2.3
wav2vec 2.0 BASE
LM=4-gram
2021.10
2.7
HuBERT BASE
LM=4-gram
2021.10
2.7
NST with LM Fusion
Supervision=Semi-super...
2020.05
3.9
DiscreteBERT
LM=4-gram
2021.10
4
Whisper-M + Llama 8B (4bit, r=4)
Train=Joint, Decode=PF...
2026.06
4.2
NST before LM Fusion
Supervision=Semi-super...
2020.05
4.3
SpeechT5
Joint CTC/Attention=true
2021.10
4.3
Whisper-M + Llama 8B (4bit, r=4)
Train=Joint, Decode=S2T
2026.06
4.8
Baseline
Joint CTC/Attention=true
2021.10
4.9
Lüscher et al.
Supervision=Supervised
2020.05
5
Whisper-M + Llama 8B (4bit, r=4)
Train=PF-S2T, Decode=P...
2026.06
5
Baseline (LAS + SpecAugment)
Supervision=Supervised...
2020.05
5.3
Hsu et al.
Supervision=Semi-super...
2020.05
5.39
SpeechT5
Joint CTC/Attention=false
2021.10
5.4
Kahn et al.
Supervision=Semi-super...
2020.05
5.41
HuBERT BASE
2021.10
5.5
Baseline
Joint CTC/Attention=false
2021.10
5.8
Whisper-M + Llama 8B (4bit, r=4)
Train=S2T, Decode=S2T
2026.06
5.9
wav2vec 2.0 BASE
2021.10
6.1
Whisper-M
Train=–, Decode=Beam
2026.06
6.5
Kahn et al.
Supervision=Supervised
2020.05
7.78
Hsu et al.
Supervision=Supervised
2020.05
14
Feedback
Search any
task
Search any
task