Nodes/ComfyUI INOUT/Schema Audio Parameter
ComfyUI Node

Schema Audio Parameter

The audio node that's a placeholder (for now)

By Liquid-XO·Created 6 months ago·Updated 5 months ago· 0
Schema Audio Parameter
  • value_in
  • value
nameparameter_name
io_kindinput
description
requiredtrue
accepted_formats
filename_prefixoutput

Let's not bury the lede: as of the current ComfyUI_SchemaNodes code, this node doesn't work. It shows up in the node menu as "Schema Audio Parameter," it has the same inout/schema category and the same base fields as its siblings - name, io_kind, description, required, accepted_formats, filename_prefix - but its class never overrides execute(). It inherits the base class method, and that method just raises NotImplementedError. Wire it up, run the workflow, and you get an error, not an audio file.

The README lists it alongside the working seven ("Input audio, output sound"), so the intent is clearly to have an audio parameter node that does for sound what SchemaImageParameter does for images - declare an audio field, load it in input mode, save it out in output mode. It just hasn't been implemented yet. It's a scaffold with a value output typed AUDIO and a value_in that accepts *, but no behavior behind it.

The inputs that matter

Since the node errors on execution, the inputs are best understood as what it will probably take when it's finished:

  • name / description - the stable identifier and machine-readable description.
  • io_kind - input / output, same role declaration as the other seven.
  • accepted_formats - presumably a metadata list of allowed audio extensions (wav, mp3, flac...).
  • filename_prefix - presumably the prefix for saved output files, mirroring the image node.
  • value_in - typed *, forced input; the eventual override slot.

You can set all of these and save the workflow, and the schema declarations will sit in the JSON like any other node. What you can't do is execute it successfully.

How it works (or doesn't)

The class definition in schema_nodes.py is five lines of configuration: output_type = "AUDIO", json_type = "audio", python_type = "AUDIO", RETURN_TYPES = ("AUDIO",), and nothing else. Every working node in the pack implements execute(); this one doesn't. That's not a config bug you can fix from the UI - it's genuinely unimplemented in the shipped source.

Installing it

It comes with the same pack as everything else.

cd ComfyUI/custom_nodes
git clone https://github.com/Liquid-XO/ComfyUI_SchemaNodes

Restart ComfyUI, or use ComfyUI Manager (search "ComfyUI SchemaNodes"). No pip dependencies, no model files - and no fix on the way via the install, because there's no missing dependency causing this. It's an unfinished feature.

What to do instead

If you need an audio parameter contract for an agent workflow today, you have a few honest options: use SchemaStringParameter to declare an audio file path as a string (which is honestly how most audio tooling exchanges files anyway), or keep watching the repo for the feature to land. If you need actual audio in the graph - loading, saving, mixing - that's a different pack entirely, and this node isn't the one to build on yet.

The one upside: because the node errors loudly instead of silently doing nothing, you won't ship a workflow that quietly drops your audio. When execute() shows up in a future version, the value output and the name/io_kind/accepted_formats/filename_prefix skeleton are already in place. Just don't plan your production pipeline around a NotImplementedError.

Categoryinout/schema

Inputs (7)

NameTypeDefaultDescription
nameSTRINGparameter_name
io_kindCOMBOinput2 options: input, output
descriptionSTRING
requiredBOOLEANtrue
accepted_formatsSTRING
filename_prefixSTRINGoutput
value_inopt*

Outputs (1)

NameTypeDescription
valueAUDIO