Audio MergeBatch To Channel

Small, single-purpose node: it takes up to four AUDIO inputs and merges them down into one AUDIO output. If you've ever ended up with several separate audio streams mid-graph - say, a batch of generated voice clips, or a few tracks you want combined instead of played one after another - this is the node that folds them into a single audio object the rest of your graph can treat normally.

How it works

There's exactly one required input, audio0, and three optional ones, audio1 through audio3. Anything you leave unplugged just isn't merged in - only audio0 is mandatory, so the node works fine with two inputs, three, or the full four. Everything you do connect gets combined into the single audio output. The pack doesn't ship a separate "split" counterpart for this node in the same category, so treat it as one-directional: batch/multiple in, one merged clip out.

Because it lives in WJNode/Audio next to AudioDuration_wan, audio_resample, and audio_scale, the practical use case is the same corner of the ecosystem: prepping audio for Wan or other audio-conditioned video models, where you typically need one clean AUDIO object feeding a single downstream node rather than several loose ones.

The inputs and outputs that matter

• audio0 (required, AUDIO) - your base track. This is the only input that must be connected.
• audio1, audio2, audio3 (all optional, AUDIO) - additional tracks to fold in. Connect as many or as few as you actually have.
• audio (output, AUDIO) - the merged result. Wire this into whatever expects a single audio clip downstream - a save node, a resample node, or an audio-conditioned sampler like the ones that read a Wan lip-sync track.

There's nothing else to configure. No mix level, no offset, no format picker in this node's own schema - it's deliberately a one-job utility, not a mixer.

Installing it

Via ComfyUI Manager: search ComfyUI-WJNodes, install, restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/807502278/ComfyUI-WJNodes.git
cd ComfyUI-WJNodes
pip install -r requirements.txt

No model downloads, no extra dependency beyond what the base pack needs.

Common issues & troubleshooting

Only one input is used and the rest are ignored. Double-check you actually wired audio1/audio2/audio3 into the node and not into something else downstream - since only audio0 is required, ComfyUI won't warn you if the optional slots are simply empty; the node will happily run and output just that one track.

Sample rate mismatches between inputs. The node's own schema doesn't expose any resample or rate-matching controls, so if your input clips were recorded or generated at different sample rates, run them through this pack's own audio_resample node first to bring them to a common rate before merging - don't assume this node normalizes that for you, because nothing in its inputs suggests it does.

Silence or a shorter-than-expected result. If one of your "connected" inputs is actually an empty or zero-length AUDIO object (a common failure mode further upstream in audio-generation graphs), it'll merge in as effectively nothing. Check the individual clips with a preview/save node before the merge if the combined output sounds wrong - it's faster to catch a bad upstream clip there than to debug it after merging.