ComfyUIAudioMass

Don't go hunting for this one in the node menu expecting to wire it into a graph - the name is half a lie. ComfyUIAudioMass doesn't process audio, return anything, or sit between two other nodes. It's a shell node whose whole job is to let jtydhr88's pack hook a complete copy of the AudioMass web editor into ComfyUI. Think of it as an embedded Audacity, not a graph operation.

Why would you care? Because ComfyUI's audio story is genuinely thin. Core gives you LoadAudio and PreviewAudio, and that's roughly it - load a file, look at a waveform, move on. The moment you're doing video-with-sound (LTX's native audio, MMAudio soundtracks on Wan clips, lip-sync chains, any music-video workflow), you need to clean up, trim, loop, or re-export audio files, and the standard answer is "open another app." This pack kills that round-trip: the editor lives at http://127.0.0.1:8188/audiomass/ and, thanks to the frontend extension, pops up inside ComfyUI itself.

How it actually works

The mechanism is the interesting part, because the Python side barely does anything. In __init__.py the pack registers two routes on ComfyUI's aiohttp server that serve a prebuilt copy of the AudioMass app (the ~8MB audiomass-ui folder, complete with WebAssembly codecs - LAME for MP3, libFLAC for FLAC - plus wavesurfer for waveforms and the RNN-based noise reduction). All audio processing happens in your browser; nothing is sent to any API and there's no model to download.

The real integration is the JavaScript extension. It does two things:

• Adds an "AudioMass" button to the top menu bar.
• Adds "Open in AudioMass" to the right-click menu on the two core audio nodes - LoadAudio (any file in its audio dropdown) and PreviewAudio (whatever's showing in its preview).

Behind the scenes it fetches the audio file, ships the raw bytes into the editor iframe via postMessage, and when you hit save it posts the edited blob back to ComfyUI's /upload/image endpoint. Then it does the clever bit: it pushes the resulting filename straight into your LoadAudio node's audio dropdown and selects it, or refreshes the PreviewAudio preview. Edit, save, run - no dragging files between folders.

The node itself: no inputs, no outputs

For real. The info_schema is empty on both sides - required: {}, optional: {}, outputs: []. The class is a stub (OUTPUT_NODE = True, returns None) that exists to register the pack. You won't add it to a workflow and you shouldn't try. The value is the button, the context-menu entries, and the URL. If someone's workflow references it, you can safely treat it as "this pack needs to be installed," not "this node needs to be wired."

Install

No Python dependencies (dependencies = []), no model files, no GPU requirements - everything ships in the repo. Either way you install it, restart ComfyUI after:

cd ComfyUI/custom_nodes
git clone https://github.com/jtydhr88/ComfyUI-AudioMass

ComfyUI Manager also finds it if you search "ComfyUI-AudioMass".

Gotchas worth knowing

• The top-menu button needs a recent frontend (ComfyUI 1.24.4+, roughly post-August 2024). If the button's missing but the pack installed fine, hit http://127.0.0.1:8188/audiomass/ directly - that always works and is the same editor.
• If that URL returns the bare text "AudioMass UI not found", the bundled audiomass-ui folder didn't come with your checkout (incomplete clone). Re-clone the full repo.
• Editing happens in-browser, so the process is local and fast, but it's browser-bound: a hard refresh wipes unsaved work, so hit save before you close the dialog.
• License: AudioMass is GPL-3.0, and the pack inherits that for the bundled editor. Fine for personal use, worth knowing if you're shipping ComfyUI to other people.