ComfyUI Node
SAM-Audio Model Loader
Loads an official SAM-Audio checkpoint or a local model from ComfyUI/models/sam_audio. If Hugging Face denies gated access, the loader uses a pinned, checksum-verified public mirror of the same checkpoint.
SAM-Audio Model Loader
- pipeline
◄model_namefacebook/sam-audio-large►
◄attention_backendpytorch►
Categoryaudio/SAM-Audio
Inputs (2)
| Name | Type | Default | Description |
|---|---|---|---|
| model_name | COMBO | facebook/sam-audio-large | Models download into models/sam_audio; valid local model folders appear automatically. |
| attention_backend | COMBO | pytorch | PyTorch SDPA is the accurate default. Comfy Kitchen uses its quantized INT8 attention kernel when supported. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| pipeline | SAM_AUDIO_PIPELINE | — |