Amplitude To Graph

This pack is all about turning audio into a curve and driving an animation with it - which means you'll constantly want to look at that curve. AmplitudeToGraph is the node for that: it takes an AMPLITUDE series (raw, from BatchAmplitudeSchedule or after ClipAmplitude/TransientAmplitudeBasic) and renders it as an IMAGE line plot you can preview. It's the "am I working with something sensible" check before you normalize and wire the curve into generation.

The plot is amplitude over frame - your loudness envelope at a glance. Wire the graph_image output into a PreviewImage node and you'll instantly see whether your frequency band picked up what you wanted (kick drum = low rumble, vocals = mid, hi-hats = shimmer at the top), and whether your curve is spiky enough that you should route it through TransientAmplitudeBasic before using it. Mechanically it's a matplotlib plot rendered to a PNG in memory and passed out as an IMAGE tensor; width scales with series length, and it disables scientific notation on the y-axis so big amplitude values don't render as 1e6 mush.

Inputs: just amplitude (AMPLITUDE). Output: graph_image (IMAGE).

Two things worth saying straight. First, it's a static render of the whole series - one image per run, not a live visualization - so treat it as a sanity check, not a tool for real-time tweaking. Second, if your graph is flat, don't blame the plot: the problem is upstream, most often a band range on BatchAmplitudeSchedule that missed your track's actual energy. Check AudioToAmplitudeGraph (the frequency-domain graph) to find where the energy lives, then come back here.

Install comes with the pack: ComfyUI Manager search ComfyUI-AudioScheduler, or git clone https://github.com/a1lazydog/ComfyUI-AudioScheduler into custom_nodes and restart. No models, CPU-only.