ComfyUI Node
MusicGen
Generate music using Meta's MusicGen model based on text prompts. Forked from comfyui-sound-lab's musicNode and modified to conform to ComfyUI's standard AUDIO format, making it compatible with core audio nodes like VAEDecodeAudio, SaveAudio, and PreviewAudio.
MusicGen
- audio
◄modelsmall►
◄prompt►
◄seconds5.0►
◄guidance_scale4.00►
◄seed0►
◄device▾►
Categoryaudio
Inputs (6)
| Name | Type | Default | Description |
|---|---|---|---|
| model | COMBO | small | Available models: Original models (most stable): - small: 300M parameters, fastest, reliable - medium: 1.5B parameters, balanced - large: 3.3B parameters, highest quality Stereo models: - stereo-small/medium/large: For spatial audio generation |
| prompt | STRING | — | |
| seconds | FLOAT | 5.01–1000 | — |
| guidance_scale | FLOAT | 4.000–20 | — |
| seed | INT | 00–2147483647 | — |
| device | COMBO | 2 options: auto, cpu |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| audio | AUDIO | — |