Nodes/ComfyUI-Mega-ASR/Mega-ASR Environment Status
ComfyUI Node

Mega-ASR Environment Status

The Mega-ASR node you run when the loader won't load

By starsFriday·Created 3 months ago·Updated 3 months ago· 0
Mega-ASR Environment Status
    • status_json

    Most ComfyUI troubleshooting is guessing. This node is the exception - it's a single-button diagnostic that tells you exactly why the Mega-ASR stack isn't running. It has no inputs and no controls; you drop it on the canvas, run it, and read the JSON it spits out.

    What it checks

    The node looks at two things, and it's ruthlessly specific about both. First, the model layout under ComfyUI/models/Mega-ASR: it reports the expected model root, whether the model files are complete (model_files_ok), and - this is the useful part - a missing_model_items list that names every missing file and folder, down to individual Qwen3-ASR weight shards referenced by the safetensors index. If you downloaded the model and a shard silently didn't come through, this is the node that tells you which one.

    Second, the Python dependencies. It reports a boolean per package - qwen_asr, safetensors, soundfile, scipy, torchaudio - so you can see at a glance whether the qwen-asr pip package actually landed in ComfyUI's Python environment instead of some other one. That's the single most common failure mode for this pack, because like every custom node, it installs into the same Python environment ComfyUI runs in, and pip doesn't care which environment you thought you were in.

    It also confirms two things about the pack itself: implementation is built_in, and requires_official_source_clone is false. That's a genuine difference from most packs - this one bundles its own inference adapter, so you do not need to clone the official Mega-ASR repo separately. If a tutorial tells you to, it's wrong.

    When to reach for it

    The README's own advice is the right order of operations: run this node first if the Loader doesn't appear in your node list, or if model loading fails. The output is a single string (status_json), so wire it into a text display node to actually read it. Expect a wall of JSON - that's the point.

    It's a boring node and it's not supposed to be exciting. It exists so you spend two minutes reading a report instead of forty minutes reinstalling dependencies that were never missing. When the report says model_files_ok is true and all the dependency flags are true, you've eliminated the environment as the problem and it's time to look at VRAM and your loader settings instead.

    CategoryAudio/Mega-ASR

    Inputs (0)

    No inputs

    Outputs (1)

    NameTypeDescription
    status_jsonSTRING