GRAudioSelector

This is GRAnySelector's audio-specific sibling - same idea, narrower type. Instead of ten wildcard slots, it's ten typed AUDIO inputs (audio_1 through audio_10), and instead of a wildcard output it's a single typed AUDIO output. The use case is exactly what you'd expect: you've got more than one possible audio source in a graph - TTS output, an uploaded track, audio pulled from a video - and you want to swap which one feeds a downstream node (a save-audio step, a video-combine node that needs an audio track) without rewiring that downstream node every time.

How it works

Wire whichever audio source you want active into one of the ten audio_ slots, and it forwards to the single AUDIO output. Like its wildcard sibling, this is a passthrough, not a mixer - it doesn't blend or layer multiple connected inputs together, and the schema doesn't expose an explicit index or priority control for choosing between several simultaneous connections. Treat it the same way: exactly one audio_ slot wired at a time, swap which one when you want to change source.

The inputs and outputs that matter

• audio_1 through audio_10 - all optional AUDIO inputs. Connect your one active source here.
• AUDIO - the single output, forwarding whatever's connected.

How to install it

ComfyUI Manager, search GraftingRayman. Manual install:

cd ComfyUI/custom_nodes
git clone https://github.com/GraftingRayman/ComfyUI_GraftingRayman

Restart afterward. The pack needs OpenAI's CLIP installed as a separate step or nothing imports - that applies here too even though this node never touches an image. Portable ComfyUI: .\python_embeded\python.exe -m pip install git+https://github.com/openai/CLIP.git. System Python: pip install git+https://github.com/openai/CLIP.git.

Common issues & troubleshooting

Pack fails to load. Check your startup console for the CLIP import error and run the pip install above.

No audio comes through at all. Confirm exactly one audio_ slot has a live connection - an unconnected slot is simply absent, not a fallback source, so if your one wired slot is actually upstream of a broken node, you'll get silence with no error pointing at this node specifically.

Behavior is unpredictable with more than one slot connected. As with GRAnySelector, there's no documented priority order for simultaneous connections in the schema. Stick to one connected audio_ slot at a time for predictable results.

You need to mix two audio sources together, not switch between them. This node doesn't do that - it's a router, not a mixer. Look for a dedicated audio-mixing node if that's the actual job.