LoadAudioVideoPath

LoadAudioVideoPath is the input half of the ComfyUI-WhisperX pack, and it exists for one reason: to hand the WhisperX node something to transcribe. You can't drag an audio file straight into the WhisperX node - it wants an AUDIOPATH, and this node is how you produce one without any Python.

It does a little more than that, though. Pick a file and the node shows you an inline audio player on the canvas so you can check you've got the right take before burning GPU on a transcription. There's also an "upload" button that copies a file straight into your ComfyUI input folder from the browser, which saves you the usual trip to the file system.

How it works

The class behind it is actually called LoadAudioPath, registered under the friendlier display name LoadAudioVideoPath. It scans ComfyUI's input directory and lists any file ending in wav, mp3, flac, m4a, or mp4 in a dropdown. Pick one, and it resolves the full path and returns it as an AUDIOPATH. The upload button and the audio preview come from a small frontend extension (uploadAudio.js) bundled with the pack, which adds a custom upload widget and rides on ComfyUI's standard /upload endpoint.

The "Video" in the name isn't decoration. It accepts mp4 because WhisperX's load_audio shells out to ffmpeg to pull out and resample the audio track - so you can point it at a rendered video and caption it directly, no separate audio extraction step.

Inputs and outputs

Just one of each, which keeps this refreshingly simple:

• audio - a dropdown of every compatible file already in your ComfyUI/input folder (the list is populated when the node is created, so add files and refresh the node to see new ones).
• Output: AUDIOPATH - wire it into the audio input of the WhisperX node.

That's the whole node. There are no models, no weights, no parameters to get wrong.

Install and gotchas

It ships with the pack, so there's nothing separate to install - if you can see WhisperX Node, you have this too. The two things that trip people up:

• "My file isn't in the list." The dropdown only shows what's already in ComfyUI/input. Either drop the file there and recreate the node, or use the node's upload button, which handles it for you. Supported extensions are wav, mp3, flac, m4a, and mp4 - anything else won't appear.
• It's a dumb pipe. If ffmpeg can't decode the file, the error doesn't show up here - it surfaces as a "Failed to load audio" RuntimeError on the WhisperX node. Don't blame this node; check the file and your ffmpeg install.

For a utility node, that's the whole story: pick a file, pass it along, hit play on the preview if you're not sure it's the right one.