List your company
ResearchFormerly WizWand

Automatic Speech Recognition on AISHELL-2 ios (dev)

2.07CERFireRedASR2S LLM
2.02442.33222.642.9478Apr 9, 2026Apr 10, 2026Apr 12, 2026Apr 14, 2026Apr 16, 2026Apr 18, 2026Apr 20, 2026
Updated 6mo ago

Evaluation results

MethodLinks
FireRedASR2S LLMModel Size=8B+, Inference Mode=Offline2026.042.07
FireRedASR-LLMModel Size=8B+ (↑)2026.042.08
Qwen3-Omni-InstModel Size=30B-A3B (↑)2026.042.11
Qwen3-Omni InstructModel Size=30B-A3B, Inference Mode=Offline2026.042.11
Step-Audio2-miniModel Size=8B+ (↑)2026.042.24
Step-Audio2 MiniModel Size=8B+, Inference Mode=Offline2026.042.24
NIM4-ASRModel Size=2.3B, Inference Mode=Offline2026.042.28
OursModel Size=2.3B2026.042.32
NIM4-ASRModel Size=2.3B, Inference Mode=Stream2026.042.33
Qwen3-ASR-1.7BModel Size=2.0B (↓)2026.042.41
Qwen3-ASR 1.7BModel Size=2.0B, Inference Mode=Offline2026.042.41
Fun-ASR-nanoModel Size=0.8B (↓)2026.042.62
Fun-ASR NanoModel Size=0.8B, Inference Mode=Offline2026.042.62
GLM-ASR-nanoModel Size=1.5B (↓)2026.043.21
GLM-ASR NanoModel Size=1.5B, Inference Mode=Offline2026.043.21