Audio Stack (mtb)

The counterpart to Audio Sequence. Where Sequence lines clips up end-to-end, Audio Stack overlays them - plays them all at the same time, mixed together. Voiceover over background music, a sound effect on top of ambience, two layers of a soundscape: this is the node that sums them into one.

It's a simple mixer, and a useful piece now that ComfyUI does audio and video. Feed it your layers, get one combined clip out.

How it works

You wire in multiple audio clips (dynamic inputs - add as many layers as you want) and the node overlays them into a single track, starting them all together. It pads the shorter clips out to the length of the longest so nothing gets cut off, resamples everything to the highest sample rate among the inputs, and promotes all of them to stereo if any single input is stereo. The result is one mixed clip.

The inputs and outputs that matter

There are no widget parameters to set - you just connect the audio layers you want to combine. The single output is stacked_audio (AUDIO), the mixed result.

How to install it

Via ComfyUI Manager: search MTB Nodes, install, restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/melMass/comfy_mtb

then restart. Like the other audio nodes it depends on ComfyUI's audio stack (torchaudio).

Common issues & troubleshooting

The mix clips / distorts. Stacking sums the signals, and this node just overlays - it doesn't automatically normalize levels. Two hot clips added together can push past full scale and clip. Bring the individual levels down before stacking, or normalize afterward.

One layer is way louder than the rest. Same root cause - there's no per-layer gain control on this node. Balance the clips' volumes upstream before they reach the stack.

The clips don't line up in time. Stack starts everything at the same moment and pads to the longest clip. If you wanted them offset - music starting, then narration coming in later - you need to arrange the timing yourself (pad or sequence a clip with silence first), because Stack aligns them all to time zero.

I actually wanted them one after another. That's Audio Sequence, not Stack.