Split Audios

ComfyUI's audio plumbing still isn't as friendly as its image plumbing. You can build a list of audio clips, but then a node downstream wants one audio at a time, and there's no convenient "index" widget to pick it. Split Audios is the blunt fix: feed it an audio list and it splits out ten separate single-audio outputs, AUDIO0 through AUDIO9. No settings, no surprises - one input, ten outputs.

When you'd actually use it

In the ComfyUI-Easy-Media world, audio lists show up all over the multitrack and timeline workflows - you build them with easy makeAudioList and the lists feed dubbing and S2V pipelines. The classic use case for the splitter: you have a fixed number of dialogue clips (say, four segments of a podcast) and you want each one routed to its own conditioning node or its own task segment without writing an index selector. You just wire AUDIO0 to the first speaker's node, AUDIO1 to the second, and so on. The 10-output limit means it's for the "I have a handful of clips" case, not for batch processing 200 files.

Honestly, it's a utility node - you'll know in thirty seconds whether you need it. If you're looping dynamically, you're better off with a list-indexing node; if your workflow has a fixed, small number of audio slots, this is the simplest possible wiring.

How it behaves

The mechanics are straightforward: the node takes the list, unwraps it, and emits each element on its own named output. Empty list positions come out as None, which is actually convenient - you can leave unused outputs unconnected and the graph won't complain. Inputs that exceed the ten slots just don't get an output, so keep lists at ten or fewer.

Install

It ships with the whole ComfyUI-Easy-Media pack:

cd ComfyUI/custom_nodes
git clone https://github.com/yolain/ComfyUI-Easy-Media.git

then restart ComfyUI (or ComfyUI Manager → "Easy-Media"). FFmpeg on your system is the pack's one flagged prerequisite, and audio work in particular leans on it.

The gotcha

The ten-output shape is the feature and the limitation. Because there's no setting to change how many outputs you see, you can't make it "split into 3" - you always get ten sockets and you ignore the extras. That's fine, but it means the node is a poor fit for variable-length audio lists. For that, look at the pack's list-side nodes or grab items by index instead. And one more thing worth knowing: AUDIO0-style outputs are single items, not lists - if a downstream node expects a list, you'll want easy makeAudioList on the other side to wrap it back up.