凤希AI - 音频管理器

Half the friction in a video workflow isn't the generation - it's babysitting your asset folders. FxAiAudioManager (凤希AI - 音频管理器) is the pack's answer: a small audio file manager that lives in a panel inside ComfyUI instead of you alt-tabbing to Explorer to rename final_final2.mp3. It manages a folder of audio files, lets you preview, upload, reorder, and delete them from the UI, and hands the workflow back the folder path plus a count of what's in it.

What it actually manages

The files live under ComfyUI/fxai/audio/<目录>. The 目录 (directory) input, default sucai, picks the subfolder - so you can keep one folder per project and swap projects by changing one string. Uploaded files are converted to wav via ffmpeg and renamed with a zero-padded index prefix (000_name.wav, 001_name.wav...), which is what makes the ordering stable for downstream nodes like FxAiAudioDurationSeg that sort by filename. You can reorder files in the panel, delete the ones you don't want, and hit apply.

It also does one genuinely useful thing that's easy to miss: if you wire an AUDIO into the optional 音频 input, it saves that generated audio straight into the folder as the next numbered wav. That turns the pack's audio generation pipeline into a library-building loop - generate a segment, save it to the folder, merge them all later with FxAiAudioMerge.

Inputs and outputs

• 目录 - string, the subfolder to manage. Defaults to sucai.
• 音频 (optional) - an AUDIO input. If connected, the node saves it to the managed folder.
• Outputs: 文件夹路径 (folder path, the absolute path you'll feed other nodes), 音频总数 (audio count), and 音频 (the audio you fed in, passed through).

One quirk worth knowing: the node is marked to always re-run (it returns NaN from IS_CHANGED), so it fires on every queue even though it doesn't strictly need to. That's the standard "always refresh the file list" trick - the trade-off being it defeats caching on that branch.

Install

One node in fxai-toolkit (凤希全能节点包), the Chinese MIT-licensed utility suite by 凤希AI (www.fxai.site) aimed at low-VRAM long-video work. Install via ComfyUI Manager (search "fxai") or:

cd ComfyUI/custom_nodes
git clone https://github.com/fxai666/fxai-toolkit
# restart ComfyUI

The pack auto-installs soundfile and psutil. Uploads are converted with ffmpeg, so that binary needs to be reachable - this node will still list files if ffmpeg is missing but conversion will fail.

Gotchas

• The panel is a browser UI - if you don't see it, hard-refresh ComfyUI (Ctrl+Shift+R) after installing so the JS loads. A stale page is the number-one "my manager node is blank" cause.
• File paths are relative to the ComfyUI install root, not wherever you keep your downloads. Drop files into ComfyUI/fxai/audio/sucai/ by hand and they'll show up too.
• Since it always re-runs, put it at the start of your chain where it won't silently force every downstream node to recompute. If your graph feels slow and you have one of these in the middle, this is why.
• Chinese labels throughout: 目录 = directory, 总数 = count.