List your company
ResearchFormerly WizWand

Automatic Speech Recognition on AISHELL-2 iOS (test)

2.27WERSeed-ASR (CN)
2.06083.47294.8856.2971Nov 29, 2022May 25, 2023Nov 18, 2023May 14, 2024Nov 7, 2024May 3, 2025Oct 28, 2025
Updated 7mo ago

Evaluation results

MethodLinks
Seed-ASR (CN)2024.072.27
Ming-Flash-Omni2025.102.44
Qwen3 Omni2025.102.55
Kimi Audio2025.102.56
Paraformer-large2024.072.85
Paraformer-largeTask-specific fine-tuning=true2023.112.9
Qwen2 Audio2025.102.92
Qwen-Audiozero-shot=true2023.113.1
Qwen-Audio2024.073.1
Hubert+Baichuan2Aishell-2 set average=true2024.073.5
MMSpeechunsup-speech=AISHELL-2, fine-tuning=true2022.113.9
MMSpeechunsup-speech=WenetSpeech, fine-tuning=true2022.113.9
MMSpeech-baseTask-specific fine-tuning=true2023.113.9
Conformer2022.115.3
MMSpeechunsup-speech=WenetSpeech, fine-tuning=false2022.116.2
MMSpeechunsup-speech=AISHELL-2, fine-tuning=false2022.116.7
Transformer2022.117.5