Automatic Speech Recognition on M4Singer
5.25CERFun-ASR Nano
Evaluation results
| Method | Links | |
|---|---|---|
| Fun-ASR NanoModel Size=0.8B, Inference Mode=Offline2026.04 | 5.25 | |
| Qwen3-ASR 1.7BModel Size=2.0B, Inference Mode=Offline2026.04 | 5.72 | |
| NIM4-ASRModel Size=2.3B, Inference Mode=Offline2026.04 | 6.39 | |
| NIM4-ASRModel Size=2.3B, Inference Mode=Stream2026.04 | 6.94 | |
| Qwen3-Omni InstructModel Size=30B-A3B, Inference Mode=Offline2026.04 | 8.4 | |
| Step-Audio2 MiniModel Size=8B+, Inference Mode=Offline2026.04 | 9.68 | |
| GLM-ASR NanoModel Size=1.5B, Inference Mode=Offline2026.04 | 18.45 |