List your company
ResearchFormerly WizWand

Speech Recognition on AISHELL-1 (dev)

1.2WERQwen-Audio
1.03082.17293.3154.4571Nov 29, 2022Mar 22, 2023Jul 13, 2023Nov 3, 2023Feb 24, 2024Jun 16, 2024Oct 7, 2024
Updated 8mo ago

Evaluation results

MethodLinks
Qwen-Audiozero-shot=true2023.111.2
MMSpeechencoder size=LARGE2022.111.6
MMSpeech-largeTask-specific fine-tuning=true2023.111.6
MMSpeechencoder size=BASE2022.112
MMSpeech-baseTask-specific fine-tuning=true2023.112
HuBERTencoder size=LARGE2022.113.1
CR-CTC, Zipformer-MParams (M)=66.22024.103.72
Wav2Vec 2.0encoder size=LARGE2022.113.8
CR-CTC, Zipformer-SParams (M)=23.12024.103.9
CTC/AED, Zipformer-MParams (M)=83.22024.104
HuBERTencoder size=BASE2022.114.1
Pruned transducer, Zipformer-MParams (M)=73.42024.104.13
CTC/AED, BranchformerParams (M)=45.42024.104.19
Wav2Vec 2.0encoder size=BASE2022.114.2
CTC/AED, E-Branchformer in ESPnetParams (M)=37.92024.104.2
Pruned transducer, Zipformer-SParams (M)=30.22024.104.4
CTC, Zipformer-MParams (M)=66.22024.104.47
CTC/AED, Zipformer-SParams (M)=39.32024.104.47
CTC/AED, Conformer in ESPnetParams (M)=46.22024.104.5
PDS-Base-8CTC=Yes2023.054.72
PDS-Base-16CTC=Yes2023.054.76
CTC, Zipformer-SParams (M)=23.12024.104.89
Stack-4CTC=Yes2023.054.96
PDS-Base-8CTC=No2023.055.09
PDS-Base-32CTC=Yes2023.055.17
PDS-Base-16CTC=No2023.055.3
Stack-4CTC=No2023.055.42
PDS-Base-32CTC=No2023.055.43