LoadAudioPath

There's nothing glamorous about LoadAudioPath, and that's fine - it's the one-line utility that feeds HalloNode its voice. HalloNode's driving_audio input needs a file path on disk, not audio loaded into memory, because the Hallo inference engine runs as a separate script that reads the audio straight from a file. This node is how you get it there.

What it does

On load it scans your ComfyUI input/ folder for audio files with wav, mp3, flac, or m4a extensions and shows them in a dropdown. Pick one, and it returns the resolved path to feed into driving_audio. The wiring is dead simple:

LoadAudioPath ──► HalloNode.driving_audio
LoadImagePath ──► HalloNode.source_image
HalloNode.Video ──► PreViewVideo

The gotcha: no upload button

Here's the thing that bites people: unlike LoadImagePath (and most core loaders), this node has no upload/drag-and-drop flag in its source. You cannot drag an audio file onto the node and have it copied into input/. You have to drop the file into the ComfyUI input/ folder yourself - via your file manager, the web UI's upload endpoint, wherever - and then it'll appear in the dropdown. Do that before you start ComfyUI if you can, because the file list is also built at load time; files added while the server is running won't show until you reload.

A couple of practical notes while you're in here:

• Name your file something sane before you load it. The dropdown shows raw filenames, and you'll be staring at this list every run.
• WAV in. The Hallo pipeline resamples to 16 kHz internally regardless of what you give it, but starting from a clean WAV avoids any codec weirdness with the vocal separator stage.
• One clip, one run. Hallo animates to the single audio you give it. There's no batch built in, so a 30-second clip means one long generation, not a bunch of short ones.

Install

This node ships in the AIFSH/ComfyUI-Hallo pack, so installing the pack gives you all three of its loaders at once: ComfyUI Manager → search ComfyUI-Hallo, or git clone https://github.com/AIFSH/ComfyUI-Hallo.git into ComfyUI/custom_nodes/ and pip install -r requirements.txt (plus ffmpeg on your PATH - Hallo needs it to mux the output). You won't install LoadAudioPath alone, and honestly you wouldn't want to; it only exists to serve HalloNode.