Automatic Speech Recognition on GigaSpeech 2 Viet
18.1Word Error Rate (WER)MiDashengLM
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| MiDashengLMModel Size=7B2025.08 | 18.1 | 18.1 | |
| Qwen2.5-OmniModel Size=7B2025.08 | 18.6 | 18.6 | |
| Kimi-Audio-InstructModel Size=7B2025.08 | 100 | 100 |