VoxCPM Save Audio

Every VoxCPM generation eventually needs to become a file you can actually listen to, and that's this node's entire job. VoxCPM Save Audio takes an AUDIO and writes it to disk as a WAV using soundfile. It's an output node, so it has no outputs - it just saves and reports where it put things.

Two inputs, both forgiving:

• audio (AUDIO) - whatever the TTS node produced.
• filename - default output.wav. Any name works, but stick to .wav; it's the format this reliably writes.
• output_dir (optional) - leave it blank and the file lands in ComfyUI's standard output folder. Give it a path and it writes there instead, creating the directory if needed.

It prints the full save path to the console and returns a little UI text ("Saved: …") so you always know where the file went. That's the whole feature list, and that's fine.

Where it fits

It's the natural end of every basic pipeline in this pack:

[VoxCPM Model Loader] → [VoxCPM Text-to-Speech] → [VoxCPM Save Audio]

Same for the cloning path - Audio Reference's output, or the concatenated result of a batch, both feed straight in. If you're writing multiple files in one run, reach for VoxCPM Batch Save Audio instead, which does the numbering for you.

Installing it

Part of the ComfyUI-DN-VoxCPM pack:

cd ComfyUI/custom_nodes/
git clone https://github.com/DekaNear/ComfyUI-DN-VoxCPM.git
cd ComfyUI-DN-VoxCPM
pip install -r requirements.txt

Restart ComfyUI, or install "DN VoxCPM" from ComfyUI Manager.

Where people get burned

The honest caveat is that it overwrites without asking. Run the same workflow twice with the default output.wav and your first render is gone. Give files distinct names (or a per-run output_dir) the moment you care about keeping something. Also worth knowing: "saved" doesn't mean "in your browser" - check ComfyUI's output folder, not your Downloads. One run, one file, no surprises after that.