AI Paper Atlas

Research map / Benchmarks / LibriSpeech test-clean

LibriSpeech test-clean leaderboard

33 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on LibriSpeech test-clean. Best WER: 1.5 (MGM-Omni-32B, ECCV 2026).

Numbers are as each paper reports them, read by AI from its results tables and not checked by hand. Papers can differ in setting (training data, model size, extra data); the topic column shows what each paper is about. Open the paper before you compare or quote a number.

LibriSpeech test-clean · WER

33 papers, each with its best reported number for its own method; lower is better.

#WER ↓MethodPaperTopic
11.5MGM-Omni-32B
Qwen2.5-VL-32B
MGM-Omni: Scaling Omni LLMs to Personalized Long-Horizon Speech
ECCV 2026 · Table 2
Language, safety and interpretability
21.57Audio Flamingo 3
Qwen2.5-7B
Audio Flamingo 3: Advancing Audio Intelligence with Fully Open Large Audio Language Models
NeurIPS 2025 · Table 1
Language, safety and interpretability
31.8Ours
WavLM Large + Vicuna-7B
Speech Recognition Meets Large Language Model: Benchmarking, Models, and Exploration
AAAI 2025 · Table 6
Language, safety and interpretability
41.81NaturalSpeech 3NaturalSpeech 3: Zero-Shot Speech Synthesis with Factorized Codec and Diffusion Models
ICML 2024 · Table: evaluation results on L
Autoregressive and latent generation
51.85Micro-batch clipping, b=4
600M Conformer XL
Revisit Micro-batch Clipping: Adaptive Data Pruning via Gradient Manipulation
ICLR 2025 · Table: Best WER on LibriSpeech
Optimizers
61.88HumanOmni-Speaker
Qwen2.5-Omni-3B
HumanOmni-Speaker: Identifying Who said What and When
ECCV 2026 · Table 4
Video-language understanding
71.88Pruned transducer with CR-CTC, Zipformer-L
Zipformer-L
CR-CTC: Consistency regularization on CTC for improved speech recognition
ICLR 2025 · Table: LibriSpeech results
Recognition and long-tail
81.9OWLS 9BOWLS: Scaling Laws for Multilingual Speech Recognition and Translation Models
ICML 2025 · Table: WER/CER of OWLS models
RNNs and normalization
92Semiring Distillation, α_state=0.001
0.6B causal Conformer RNN-T
Revisiting the Entropy Semiring for Neural Speech Recognition
ICLR 2023 · Table: Self-Distillation for a
Knowledge distillation
102.06Zipformer-LZipformer: A faster and better encoder for automatic speech recognition
ICLR 2024 · Table 1
Attention and transformers
23 more papers on this leaderboard, the baselines reported on it and a CSV download are free with a scime account.
Sign in free to see all

Not affiliated with any conference, journal or the benchmark's authors. Corrections: contact@latenticlabs.com.