List To Audio

Audio in ComfyUI doesn't come as a single thing the way an image does. Plenty of video nodes hand you a list of AUDIO objects - one per segment, one per shot, one per generated chunk - and the moment a list shows up, every downstream node that wants a plain AUDIO input stops talking to you. List To Audio is the fix: it swallows a list of audios and returns a single AUDIO.

Don't misread the name, because the difference matters. This is not a mixer or a concatenator. It won't blend two clips together or join them end to end. Under the hood it's a list-unwrapper: the node declares INPUT_IS_LIST = True, so it receives the whole list, and then simply returns audios[0] - the first element. That's the entire mechanism. If the list happens to be empty, it returns a silent, zero-length audio at 16kHz instead of crashing, which is a nice defensive touch you'll appreciate when an upstream node legitimately produced nothing.

This pattern should look familiar if you've spent time in the plumbing layer of ComfyUI: it's the same "unwrap a single-element list" idiom you see everywhere in utility packs, just specialized for the audio type. The pack author's sibling repo, ComfyUI-ListUnwrap, does this for the old Director combine workflows, and this node is that same idea shipped inside the Chain Director pack so you get it for free. In a MiniMax H3 context it's the bridge between a node that outputs several segments' audio as a list and a consumer (like a VHS_VideoCombine or an audio preview) that demands exactly one.

The inputs and output are as minimal as it gets: audios (AUDIO, as a list) in, a single AUDIO out. There's nothing to configure. The only real decision is upstream: since it returns the first element, make sure the list is ordered the way you want, and that you're not accidentally dropping the audio you cared about. If a node gives you three segment audios and you want all of them in one track, this isn't the node for that - you want a concat node, or the Chain Director itself, which concatenates segment waveforms internally (torch.cat along the time axis) before handing you the final audio.

Install is the shared pack install: ComfyUI Manager search "ChainDirector", or git clone https://github.com/luxu1999/ComfyUI-MiniMaxH3-ChainDirector.git into custom_nodes/, then restart ComfyUI. No extra dependencies. It's a five-line node that does one tiny job, and if you've ever fought a list-of-audio error at 2am, you'll be glad it exists.