vlo Memory Load Audio

Audio is the awkward kid in ComfyUI. There's no built-in audio upload widget the way images have, and once video models started generating synchronized soundtracks (looking at you, LTX-2), the "how do I get a wav into this graph" problem got loud. vlo Memory Load Audio is one answer: an audio loader that pulls from an in-memory registry by id, no file on disk required.

It's the audio sibling of the vlo memory loaders, and it exists for the same reason - vlo, the local open-source video editor these nodes bridge to, hands media to ComfyUI over HTTP rather than dropping files into input. vlo's timeline deals in audio clips: reference tracks, foley, dialogue. When you send one of those into a generation, this node is what picks it up on the Comfy side, decoded from the registry bytes into a proper ComfyUI AUDIO payload (waveform tensor plus sample rate) by PyAV.

Inputs

• audio - a dropdown listing registry ids for audio (and video files, since many videos carry a usable audio track). The refresh button re-pulls the current list from vlo.
• disable_in_memory (boolean, default false) - same escape hatch as the other loaders: flip it and the dropdown lists actual files from your ComfyUI/input folder, so you can hand-test a workflow without vlo running.

One output: AUDIO - the standard ComfyUI audio bundle (waveform, sample_rate) that LTX audio-conditioned workflows and audio nodes expect.

The honest details

Two things are worth knowing. First, the audio dropdown is deliberately not a native upload widget - the pack keeps it a remote-backed combo because ComfyUI's built-in audio upload preview can crash the frontend for custom node classes. If you're tempted to wire your own audio uploader in, you're fighting the pack's design. Second, the registry rules are shared across all the memory loaders: 512 MB per item, 2 GB total, unread items expire after two hours, and everything is lost when ComfyUI restarts. A stale id error ("Unknown media id") means re-send the audio from vlo - it's not on your disk to check.

Install

Part of ComfyUI-vlo, so:

cd ComfyUI/custom_nodes
git clone https://github.com/PxTicks/ComfyUI-vlo.git

Restart ComfyUI (or install via ComfyUI Manager as "ComfyUI-vlo"). The web extension registers itself. No models to download; PyAV for decoding ships with modern ComfyUI, so there's nothing extra to pip install on a current setup.