Load Audio

This is the one-trick-pony of the ComfyUI-JM-MiniMax-API pack, and that's fine - the trick is one you can't easily do without it. Load Audio exists to hand a real file path to the Voice Cloning node, which is the one node in this pack that needs to upload a local file to MiniMax's servers.

If you've used ComfyUI's stock Load Image node, this works the same way, just for audio. It shows you a dropdown of audio files sitting in ComfyUI's input directory, you pick one (or upload one through the button), and it outputs an absolute path. That's the entire node - no processing, no preview magic, no analysis. It's a path-passing node, and that's exactly what the next node needs.

Why it has to exist

Voice Cloning's audio_file input expects a real filesystem path to an audio file it can open and upload. ComfyUI has no built-in audio picker, so this pack shipped its own: a dropdown of .mp3, .wav, and .m4a files it finds in the input directory, plus a custom upload widget.

That upload widget is the quietly clever bit. The pack registers a web extension (web/jm_audio_upload.js) with ComfyUI's frontend, which is what makes the Upload button appear on the node. Click it, pick a file, and it lands straight in your input directory - no manual copy-paste into the folder needed. It supports the same three formats as the dropdown.

Inputs and output

• audio_path - a dropdown of audio files in the input directory. This is the one you actually use.
• upload - the optional custom widget that lets you add new files without touching the filesystem.

The output is audio_file: the absolute path to the selected file as a string. Wire it into the Voice Cloning node's audio_file input.

There's also some subtle behavior worth knowing: the node tracks the file's modification time, so if you replace the audio file with the same name, ComfyUI will notice and re-run the downstream nodes. Drop a new sample in, run again, get a fresh clone.

Gotchas

The dropdown only lists files already in ComfyUI's input directory (matching .mp3, .wav, .m4a). If your file isn't showing up, that's where it needs to go - or just use the upload button, which puts it there for you.

And remember it's a local path, not an actual audio analysis tool. If you need to trim, denoise, or check levels before cloning, do that outside ComfyUI (or with an audio node pack) first. Garbage in, mediocre clone out.

Install

It ships inside the ComfyUI-JM-MiniMax-API pack, so one install gets all eight nodes:

cd ComfyUI/custom_nodes
git clone https://github.com/synthetai/ComfyUI-JM-MiniMax-API
pip install -r requirements.txt

Or search "ComfyUI-JM-MiniMax-API" in ComfyUI Manager and restart. Requirements are just requests and Pillow - and yes, for a node that never makes a network call, this one still ships in an API pack. It's the front door for the voice-cloning workflow that follows.