VRGDG_LoadAudioWithPath

Most audio loaders in this pack give you either a file path or a loaded clip. VRGDG_LoadAudioWithPath gives you all three at once: the decoded AUDIO, the full file path, and a cleaned-up filename. It's the node you put at the front of a splitter pipeline when you want the track and its provenance in one place.

The input is a plain file dropdown - you pick the audio from a list (audio, an enum of available files). That's the whole control surface, which makes this the easiest loader to wire up: there's no path typing, no refresh knob to remember.

What it does

One input, three outputs:

• audio (enum) - the file picker.
• audio (AUDIO, output) - the decoded track. Wire this into any splitter that takes an AUDIO input: VRGDG_LoadAudioSplit_HUMO, the Transcribe variants, the General splitter, or VRGDG_LoadAudioSplit_SRTOnly.
• audio_file_path (STRING, output) - the full path. Feed this to anything that wants a path, like VRGDG_LoadAudioSplitDynamic or a save node.
• audio_name_clean (STRING, output) - the filename with directory cruft stripped (and, depending on the source, formatting normalized). Perfect for naming outputs after the track.

The one-input-to-three-outputs shape is what makes it compose well: one node at the front of the graph, and every downstream branch gets the flavor of data it needs. The splitters need the decoded audio; the LLM branch may want the track name for prompt context ("this is a song called X"); save nodes want the path. This node satisfies all three without you wiring three separate loaders.

Why "clean name" earns its place

File names are where pipelines go to die. Tracks arrive as song_ft_artist_FINAL_v3 (1).mp3 or worse, and if that string leaks into an output filename or an LLM prompt, you get nonsense downstream. audio_name_clean is the pack's way of saying "we know the source names are gross, here's the tidy version." Use it for output naming and you'll thank yourself when you're hunting for the right render later.

Installing it

Part of the comfyui-vrgamedevgirl pack. ComfyUI Manager: search "vrgamedev", install, restart. Or:

cd ComfyUI/custom_nodes
git clone https://github.com/vrgamegirl19/comfyui-vrgamedevgirl

then install the README's requirements (librosa handles the decode here):

pip install -r custom_nodes/comfyui-vrgamedevgirl/requirements.txt

If it's not working

If the file dropdown is empty, there's no audio in the directory the pack scans - drop a file in and refresh the node. If the loaded audio is silent or truncated, the source file itself is likely the problem (try a clean export). And if audio_name_clean comes back with more cruft than you expected, don't fight it - that output is best-effort; you can always regex the name further downstream.