AI Paper Atlas

Research map / Benchmarks / LibriSpeech

LibriSpeech leaderboard

11 papers at ICML, NeurIPS, ICLR, CVPR, AAAI, ICCV and ECCV, and the journals TPAMI, IJCV and JMLR report results on LibriSpeech. Best WER: 1.72 (HRF-C(3,3), ICLR 2022).

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 · WER

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

#WER ↓MethodPaperTopic
11.72HRF-C(3,3)
17-layer Conformer-Transducer
Hybrid Random Features
ICLR 2022 · Table, Training Speech Models
Kernels and regression theory
21.99MReQ-SpeechTokenizer
SpeechTokenizer
HALL-E: Hierarchical Neural Codec Language Model for Minute-Long Zero-Shot Text-to-Speech Synthesis
ICLR 2025 · Speech reconstruction table
Language, safety and interpretability
32.26NaturalSpeech 2NaturalSpeech 2: Latent Diffusion Models are Natural and Zero-Shot Speech and Singing Synthesizers
ICLR 2024 · Word error rate table
Autoregressive and latent generation
42.54SimulMEGA-TTS
CosyVoice 2
SimulMEGA: MoE Routers are Advanced Policy Makers for Simultaneous Speech Translation
NeurIPS 2025 · Table (TTS evaluation results)
Mixture of experts
52.79VLAS-Base
Vicuna-7B
VLAS: Vision-Language-Action Model with Speech Instructions for Customized Robot Manipulation
ICLR 2025 · Table: Performance comparison
Robot action and navigation
62.8BEST-RQ, non-streaming pre-train
Conformer 0.6B
Self-supervised learning with random-projection quantizer for speech recognition
ICML 2022 · Table 2
Self-supervised and contrastive
74.63SECodecSECodec: Structural Entropy-based Compressive Speech Representation Codec for Speech Language Models
AAAI 2025 · Table 1
Graph self-supervised learning
85.04SpeechTokenizerSpeechTokenizer: Unified Speech Tokenizer for Speech Language Models
ICLR 2024 · Speech reconstruction table
Language, safety and interpretability
96.24Ours, 50HzScaling Speech-Text Pre-training with Synthetic Interleaved Data
ICLR 2025 · Speech Reconstruction Results
Language, safety and interpretability
107.95Sylber 20K
SDHuBERT
Sylber: Syllabic Embedding Representation of Speech from Raw Audio
ICLR 2025 · Resynthesis table
Self-supervised and contrastive
1 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.