KokoroTTS_LoadVoice_v1

This is a small, mechanical node with one job: read a saved custom voice file off disk and hand it to whatever downstream node needs a VOICE value. It's not glamorous, but it's the hinge the whole custom-voice half of this pack turns on - without it, a voice you blended and saved has nowhere to go.

Where it fits

The pack's save/load pair works like this: KokoroTTS_SaveVoice_v1 blends two of Kokoro's presets and writes the result to a .kkv file on disk, but its own output is just a saved_path string - a confirmation the save worked, not something a TTS node can generate speech from. KokoroTTS_LoadVoice_v1 is the other half: it reads that file back in and outputs a voice_shape, typed VOICE, that plugs straight into KokoroTTS_v4 (single custom-voice generation) or KokoroTTS_v5 (blending two custom voices together). Save writes it, Load reads it back - you need both, in that order, and neither one on its own does anything with sound.

Inputs and outputs

Just one input:

• voice_file (enum) - a dropdown of .kkv files it finds on disk. The schema's sample shows custom_voice.kkv, which happens to be the default save_name KokoroTTS_SaveVoice_v1 writes if you don't change it.

One output:

• voice_shape (VOICE) - wire this into KokoroTTS_v4's voice input, or either voice1/voice2 slot on KokoroTTS_v5.

The dropdown is populated dynamically from what's actually saved, so the natural order matters here: run SaveVoice_v1 first to create at least one .kkv file, then add this node - otherwise there's nothing in the list to pick.

Installing it

Same install as every node in the pack:

• ComfyUI Manager - search "DJZ-KokoroTTS", install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/MushroomFleet/DJZ-KokoroTTS, then restart.

Windows portable users: run the repo's install-portable.bat with ComfyUI closed first - it wires up the dependencies and paths a portable build needs.

You'll also need two model files sitting in /comfyui/models/kokoro/ for the pack to work at all - the README gives a direct link to the ONNX weights (kokoro.onnx, from taylorchu/kokoro-onnx's v0.2.0 release) but is vague about the second file's exact name, so check the repo's models.json.

Where people get stuck

The most common issue with this specific node isn't the shared model files - it's an empty or unexpected dropdown. If voice_file shows nothing to pick, you haven't saved a voice yet; go run KokoroTTS_SaveVoice_v1 first. If you have saved one and it's still not showing up, restart ComfyUI - dropdown option lists built from files on disk sometimes need a restart to refresh, the same way newly installed nodes do.

Beyond that, this pack shares the general custom-node caveat: its requirements.txt installs into ComfyUI's shared Python environment alongside everything else you've got, with no isolation between packs. If the node itself won't load - not just an empty dropdown, but missing entirely - check the ComfyUI console at startup for a dependency conflict before assuming it's a file-path problem. And since this is a small, fairly new pack, don't expect a deep well of community threads if you hit something unusual; the repo's own issues page is the better bet.