HyperAI
StartseiteNeuigkeitenNeueste ForschungsarbeitenTutorialsDatensätzeWikiSOTALLM-ModelleGPU-RanglisteVeranstaltungen
Suche
Über
HyperAI

Speech Recognition On Common Voice German

Metriken

Test WER

Ergebnisse

Leistungsergebnisse verschiedener Modelle zu diesem Benchmark

Modellname
Test WER
Paper TitleRepository
ConformerCTC-L (no LM)6.68%NeMo: a toolkit for building AI applications using Neural Modules
wav2vec 2.0 XLS-R 1B + TEVR (5-gram)3.64%TEVR: Improving Speech Recognition by Token Entropy Variance Reduction
Whisper (Large v2)6.4%Robust Speech Recognition via Large-Scale Weak Supervision
wav2vec 2.0 XLS-R 1B + TEVR (4-gram)3.70%TEVR: Improving Speech Recognition by Token Entropy Variance Reduction
wav2vec 2.0 XLS-R 1B (5-gram)4.38%TEVR: Improving Speech Recognition by Token Entropy Variance Reduction
Conformer Transducer (no LM)6.28%Automatic Speech Recognition in German: A Detailed Error Analysis-
QuartzNet15x5DE (D37, 5-gram)6.6%Scribosermo: Fast Speech-to-Text models for German and other Languages
VoxPopuli (n-gram)7.8%VoxPopuli: A Large-Scale Multilingual Speech Corpus for Representation Learning, Semi-Supervised Learning and Interpretation
ConformerCTC-L (4-gram)6.03%NeMo: a toolkit for building AI applications using Neural Modules
wav2vec 2.0 XLS-R (no LM)12.06%TEVR: Improving Speech Recognition by Token Entropy Variance Reduction
ConformerCTC-L (5-gram)4.05%Scribosermo: Fast Speech-to-Text models for German and other Languages
QuartzNet15x5DE (CV-only, 5-gram)7.7%Scribosermo: Fast Speech-to-Text models for German and other Languages
wav2vec 2.0 XLS-R 1B + TEVR (no LM)10.10%TEVR: Improving Speech Recognition by Token Entropy Variance Reduction
ConformerCTC-L (no LM)7.33%Scribosermo: Fast Speech-to-Text models for German and other Languages
0 of 14 row(s) selected.