The default workflow using the official workflow template "stable audio 3" generates audio that is entirely static.

### Custom Node Testing

- [x] I have tried disabling custom nodes and the issue persists (see [how to disable custom nodes](https://docs.comfy.org/troubleshooting/custom-node-issues#step-1%3A-test-with-all-custom-nodes-disabled) if you need help)

### Expected Behavior

This is the audio file I generated using this workflow; it's all noise.

### Actual Behavior

[stable_audio_3_00007_.mp3](https://github.com/user-attachments/files/28296266/stable_audio_3_00007_.mp3)

### Steps to Reproduce

After downloading the model, click "Run," and the generated audio will look like this.

### Debug Logs

```powershell
[INFO] got prompt
[INFO] model weight dtype torch.float16, manual cast: None
[INFO] model_type FLOW
[INFO] VAE load device: cuda:0, offload device: cpu, dtype: torch.bfloat16
[WARNING] no CLIP/text encoder weights in checkpoint, the text encoder model will not be loaded.
FETCH ComfyRegistry Data: 135/148
[INFO] CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16
[INFO] Requested to load SAT5GemmaModel
[INFO] Model SAT5GemmaModel prepared for dynamic VRAM loading. 537MB Staged. 0 patches attached. Force pre-loaded 49 weights: 73 KB.
[INFO] CLIP/text encoder model load device: cuda:0, offload device: cpu, current: cpu, dtype: torch.float16
[INFO] Requested to load Qwen35TEModel_
[INFO] Model Qwen35TEModel_ prepared for dynamic VRAM loading. 4221MB Staged. 0 patches attached. Force pre-loaded 177 weights: 402 KB.
Generating tokens: 11%|███████ | 29/256 [00:02<00:09, 24.73it/s]FETCH ComfyRegistry Data: 140/148
Generating tokens: 18%|███████████▏ | 46/256 [00:03<00:14, 14.51it/s]
[INFO] Model SAT5GemmaModel prepared for dynamic VRAM loading. 537MB Staged. 0 patches attached. Force pre-loaded 49 weights: 73 KB.
[INFO] Requested to load StableAudio3
[INFO] Model StableAudio3 prepared for dynamic VRAM loading. 2771MB Staged. 0 patches attached. Force pre-loaded 168 weights: 231 KB.
88%|█████████████████████████████████████████████████████████████████████████▌ | 7/8 [00:00<00:00, 13.25it/s]FETCH ComfyRegistry Data: 145/148
100%|████████████████████████████████████████████████████████████████████████████████████| 8/8 [00:00<00:00, 9.81it/s]
[INFO] Requested to load SA3AudioVAE
[INFO] 0 models unloaded.
[INFO] Model SA3AudioVAE prepared for dynamic VRAM loading. 3243MB Staged. 0 patches attached. Force pre-loaded 296 weights: 3377 KB.
FETCH ComfyRegistry Data [DONE]
[INFO] [ComfyUI-Manager] default cache updated: https://api.comfy.org/nodes
FETCH DATA from: https://raw.githubusercontent.com/ltdrdata/ComfyUI-Manager/main/custom-node-list.json [DONE]
[INFO] [ComfyUI-Manager] All startup tasks have been completed.
[INFO] Prompt executed in 16.54 seconds
```

### Other

_No response_