XAudioGet

XAudioGet is one of the nodes in this pack that makes no sense in isolation. It exists to be the receiving end of the pack's companion web panel, XDataHub - the sidecar app where you push media into a running ComfyUI session. When you send an audio file from XDataHub, this node is how the graph actually gets its hands on it. If you never use XDataHub, you'll never need this node, and that's fine.

How it works

The only input is media_ref, a hidden string that XDataHub fills in for you. Think of it as an opaque ticket: it doesn't name a file path you type, it's an identifier the panel resolves against its own registry. An empty media_ref means no output - the node returns nothing and moves on. When the ref is valid, the node resolves it to a file, loads the audio with PyAV, and emits a standard ComfyUI AUDIO output - a tensor plus the sample rate - that you can wire into any audio-processing or audio-save node.

There's a nice touch under the hood: fingerprint_inputs keys off the media_ref, so ComfyUI won't re-execute the load unless the ref actually changes. You can re-run the workflow, and the audio loader stays cached instead of re-reading the file every pass. It also fails soft - a missing file or an invalid ref just logs a warning and outputs nothing, rather than nuking the whole queue.

Inputs and outputs

• media_ref (STRING, hidden) - the XDataHub media reference. Leave it alone; the panel manages it. Empty means no output.
• audio (AUDIO) - the loaded audio, ready for downstream audio nodes.

That's the whole node. One string in, one audio out.

Installing it

XAudioGet is part of ComfyUI-Xz3r0-Nodes. Install the pack once (ComfyUI Manager: search ComfyUI-Xz3r0-Nodes), or:

cd ComfyUI/custom_nodes
git clone https://github.com/Xz3r0-M/ComfyUI-Xz3r0-Nodes.git
cd ComfyUI-Xz3r0-Nodes
pip install -r requirements.txt

Then restart ComfyUI. One dependency worth calling out before you build an audio workflow: the pack's README is explicit that system FFmpeg must be installed and on your PATH for the audio/video save nodes and XDataHub's video thumbnail previews. Audio loading here goes through PyAV, which also leans on FFmpeg libraries, so on Windows you're best off installing FFmpeg properly and adding it to PATH rather than hoping a bundled copy appears.

Gotchas

The number one confusion with XAudioGet is expecting it to be a file loader. It is not - there's no filename box, no folder scan. Without the XDataHub panel running and pushing audio, this node has nothing to give you. So the workflow is: open the XDataHub panel, send audio, and the node lights up with the latest send. If you want a plain "load an audio file from disk" node, this isn't it; look for a dedicated audio loader instead. And since the media ref is session-bound, don't expect the audio to persist across a fresh restart unless you re-send it.