🎵📊 Audio Visualizer

Feed it an audio file, pick a visual style, and this node turns the waveform into an actual animated image or video - waveform lines, circular waveforms, bouncing bars, particle fields, or an "umbrella" style, with the option to write your own visualizer script if none of the built-ins fit. It's one of the more purely fun nodes in this pack, and the click-through numbers back that up: a small number of people search for it, but almost everyone who finds it clicks - which tracks with it being exactly the kind of thing people go looking for by name once they know it exists.

The mode that matters most is output_type, and picking the right one changes how you should wire everything else. image gives you a single frame - pair it with seed set to 0 and ComfyUI's Control After Generate set to Increment, and each queue run advances one frame further, letting you build an animation across many runs rather than one giant render. image_batch renders every frame at once as a stack of separate images - meant to feed into the VHS (Video Helper Suite) Video Combine node alongside the original audio, not into a Preview Image node; the tooltip is direct about this, warning that previewing a large frame count can freeze or stutter ComfyUI outright. video renders a single animated video file directly, meant for ComfyUI's native Save Video node.

Other inputs worth understanding: visualizer_script picks the style - and the tooltip flags that this is a real extension point, not just a dropdown of fixed presets: drop your own Python script into /nodes/audio_visualizers/ and it shows up as an option here, which is an unusually open door for a node aimed at beginners. scale controls how responsive the visualization is to the audio's amplitude - turn it up if your source audio is quiet and the visualization looks flat. stereo_to_mono decides how a stereo track collapses to the single channel the visualizer actually processes - mean (average both channels) is the sane default, left/right if you specifically want one channel's content. framerate matters most for the video and image_batch outputs, and needs to match whatever framerate you set on a downstream VHS Video Combine node or your audio will drift out of sync with the visuals. width/height set output resolution the normal way.

seed does double duty here beyond the frame-cursor trick above: setting it to 0 also resets the visualizer's internal animation state, which matters for particle-style visualizers that carry momentum between frames. That's a deliberate design choice, not a side effect - the tooltip states it directly - but it means if you're doing the frame-by-frame single-image workflow described above, only your very first run should actually be seed 0; hitting it again mid-sequence will restart your animation from scratch rather than continuing it.

Installing it: ComfyUI Manager, search "ComfyUI-mnemic-nodes," or git clone https://github.com/MNeMoNiCuZ/ComfyUI-mnemic-nodes into custom_nodes, then restart.

Where it goes wrong: a framerate mismatch between this node and VHS Video Combine is the most common one, and it's entirely on you to keep the two numbers in sync since neither node checks the other's setting. Piping image_batch straight into a Preview Image node without thinking about frame count is the other classic - for anything more than a few seconds of audio at a reasonable framerate, that's a lot of images to render into a preview widget at once.