Automatic Speech Recognition on WenetSpeech-Yue short
5.05CERFireRedASR2S LLM
Evaluation results
| Method | Links | |
|---|---|---|
| FireRedASR2S LLMModel Size=8B+, Inference Mode=Offline2026.04 | 5.05 | |
| NIM4-ASRModel Size=2.3B, Inference Mode=Offline2026.04 | 5.12 | |
| OursModel Size=2.3B2026.04 | 5.22 | |
| NIM4-ASRModel Size=2.3B, Inference Mode=Stream2026.04 | 5.39 | |
| Qwen3-ASR-1.7BModel Size=2.0B (↓)2026.04 | 5.79 | |
| Qwen3-ASR 1.7BModel Size=2.0B, Inference Mode=Offline2026.04 | 5.79 | |
| Qwen3-ASR-1.7BParameters=1.7B2026.01 | 5.82 | |
| Qwen3-Omni-InstModel Size=30B-A3B (↑)2026.04 | 6.97 | |
| Qwen3-Omni InstructModel Size=30B-A3B, Inference Mode=Offline2026.04 | 6.97 | |
| Fun-ASR-nanoModel Size=0.8B (↓)2026.04 | 7.31 | |
| Fun-ASR NanoModel Size=0.8B, Inference Mode=Offline2026.04 | 7.31 | |
| Qwen3-ASR-0.6BParameters=0.6B2026.01 | 7.54 | |
| Step-Audio2-miniModel Size=8B+ (↑)2026.04 | 7.78 | |
| Step-Audio2 MiniModel Size=8B+, Inference Mode=Offline2026.04 | 7.78 | |
| Doubao-ASR2026.01 | 9.74 | |
| GPT-4o-Transcribe2026.01 | 15.62 | |
| GLM-ASR-nanoModel Size=1.5B (↓)2026.04 | 16.78 | |
| GLM-ASR NanoModel Size=1.5B, Inference Mode=Offline2026.04 | 16.78 | |
| Gemini-2.5-Pro2026.01 | 25.19 | |
| Whisper-large-v32026.01 | 32.26 |