Sentence Stress Reasoning on Stress (test)
92.6SSR AccuracyHuman Annotators
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Human AnnotatorsCategory=Human, Trans.=False, Stress=False, Audio=True2025.05 | 92.6 | 96 | |
| gpt-4oCategory=Text LLM, Trans.=True, Stress=True, Audio=False2025.05 | 86.2 | — | |
| StresSLMCategory=SLM, Trans.=False, Stress=False, Audio=True2025.05 | 86.2 | — | |
| WhiStress → gpt-4oCategory=Cascade Pipeline, Trans.=False, Stress=False, Audio=True2025.05 | 83.4 | — | |
| gpt-4o-miniCategory=Text LLM, Trans.=True, Stress=True, Audio=False2025.05 | 79.3 | — | |
| Gemini-2.5-ProCategory=SLM, Trans.=False, Stress=False, Audio=True2025.05 | 77.5 | — | |
| WhiStress → gpt-4o-miniCategory=Cascade Pipeline, Trans.=False, Stress=False, Audio=True2025.05 | 76.1 | — | |
| Llama-3.1-8B-InstructCategory=Text LLM, Trans.=True, Stress=True, Audio=False2025.05 | 73.3 | — | |
| gpt-4o-audioCategory=SLM, Trans.=False, Stress=False, Audio=True2025.05 | 68.8 | — | |
| Qwen2-7B-InstructCategory=Text LLM, Trans.=True, Stress=True, Audio=False2025.05 | 67.8 | — | |
| WhiStress → Llama-3.1-8B-InstructCategory=Cascade Pipeline, Trans.=False, Stress=False, Audio=True2025.05 | 65.5 | — | |
| WhiStress → Qwen2-7B-InstructCategory=Cascade Pipeline, Trans.=False, Stress=False, Audio=True2025.05 | 65.5 | — | |
| Qwen3-Omni-30B-InstructCategory=SLM, Trans.=False, Stress=False, Audio=True2025.05 | 64.6 | — | |
| Qwen-7B-ChatCategory=Text LLM, Trans.=True, Stress=True, Audio=False2025.05 | 61.4 | — | |
| Audio-Flamingo-3Category=SLM, Trans.=False, Stress=False, Audio=True2025.05 | 56.8 | — | |
| SALMONNCategory=SLM, Trans.=False, Stress=False, Audio=True2025.05 | 55.9 | — | |
| Qwen2Audio-7B-InstructCategory=SLM, Trans.=False, Stress=False, Audio=True2025.05 | 53.2 | — | |
| Phi-4-multimodal-instructCategory=SLM, Trans.=False, Stress=False, Audio=True2025.05 | 52.2 | — | |
| WhiStress → Qwen-7B-ChatCategory=Cascade Pipeline, Trans.=False, Stress=False, Audio=True2025.05 | 50.9 | — | |
| LLaMA-OmniCategory=SLM, Trans.=False, Stress=False, Audio=True2025.05 | 49.5 | — |