FlashSR Processing Settings

When a workflow has the same filter settings living in two or three separate node widgets, they will drift apart - you'll tune the PRE low-pass in one node, forget the other, and wonder why your render sounds different from yesterday. FlashSR Processing Settings exists to make that impossible. It's a pure configuration node: you set the PRE and POST low-pass values once, here, and its outputs wire directly into the override inputs of the FlashSR Lowpass Lab nodes (and the FlashSR node itself) so every stage reads the same effective settings. It also emits JSON receipts so the exact configuration lands in your metadata sidecar.

Don't expect any audio to pass through this node - there's no AUDIO input or output. It's the pack's "control panel," and it's the node you actually tune while the filter nodes become dumb slaves of their override inputs.

What the inputs do

The widget set mirrors what you'd set on a FlashSR Lowpass Lab, but twice: a pre_* block and a post_* block, each with _preset, _custom_cutoff_hz, _custom_order, _custom_phase, and _bypass.

The defaults are worth respecting. PRE defaults to PRE 12 kHz - recommended, the stated starting point for a 32 kHz MiniMax source, with zero_phase (avoid phase rotation before enhancement). POST defaults to POST 19 kHz - slightly stronger with causal phase - a gentle one-way roll-off to hide artificial air. The phase defaults differ intentionally: pre-filtering prefers zero-phase so FlashSR sees an unrotated spectrum; post-filtering defaults to causal for a natural roll-off.

There's one flag that trips people up: flashsr_lowpass_input. This passes a lowpass_input switch to the FlashSR node itself. Keep it OFF when you're already doing an explicit PRE low-pass in the chain - the tooltip is blunt about it: "enabling both can apply unintended extra filtering." Double-filtering is the classic silent mistake this node was built to prevent.

Outputs and wiring

It emits 13 outputs: pre_preset, pre_cutoff_hz, pre_order, pre_phase, pre_bypass, pre_settings_json, and the same six for POST, plus flashsr_lowpass_input. In the example workflow, the *_preset/*_bypass outputs drive the matching override inputs on your FlashSR Lowpass Lab instances, and the two settings_json outputs feed MiniMax Song Metadata. You don't have to wire every output - connecting just the ones your graph needs is fine.

Installing it

Standard pack install via ComfyUI Manager (search "MiniMax Music Production Toolkit") or:

cd ComfyUI/custom_nodes
git clone https://github.com/jplenio/ComfyUI-MiniMax-Music-Production-Toolkit.git
cd ComfyUI-MiniMax-Music-Production-Toolkit
python -m pip install -r requirements.txt

Uses your ComfyUI Python. Dependencies: scipy, soundfile, imageio-ffmpeg, mutagen, Pillow. No model weights. Restart and hard-refresh.

Gotchas

The big one is the flashsr_lowpass_input double-filter trap described above - leave it off unless you've deliberately skipped the explicit PRE low-pass. Second, remember this node only configures: if you wire it to nothing, it does nothing but sit there looking authoritative. And if your FlashSR node is missing entirely, that's the separate Egregora audio super-resolution pack this toolkit expects you to install - this node can't conjure FlashSR on its own.