♻️ Refresh Voice Cache

Small node, specific annoyance. The suite caches the list of available voice files so it doesn't scan your voices/ folder on every generation. Great for speed - irritating the moment you add a new voice .wav (or save one with the Voice Designer) and the dropdowns don't see it. The usual fix is restarting ComfyUI, which is a heavy hammer for a tiny problem. This node is the light hammer: it busts the voice cache mid-graph so freshly-added voices show up without a full restart.

It's a pass-through utility. It doesn't generate or transform audio - it sits inline in your graph, and when it runs, it forces the voice cache to refresh.

How it works

The node takes a signal input and passes it straight back out, and while doing so, refreshes the voice cache. The signal wiring is the mechanism: because ComfyUI only executes a node when something depends on its output, you thread this node's signal through a connection that's part of your run, which guarantees the refresh actually fires at the right moment rather than being skipped. Think of it as a "and also, re-scan the voices folder" step you splice into the flow.

The inputs and outputs that matter

• signal - required, a wildcard pass-through. Feed it any value/connection that's part of your graph's execution; it comes back out unchanged on the signal output. Its only purpose is to place the refresh in the execution path.
• force_refresh (default on) - actually perform the refresh. On is what you want; it's the whole reason the node exists.
• signal2 - an optional second pass-through, for when you need to route two things through the same refresh point.

Outputs mirror the inputs: signal and signal2 pass through untouched, and info reports what the refresh did (e.g. how many voices it now sees).

Installing it

Ships with the pack. ComfyUI Manager → search "TTS Audio Suite" → install → restart. Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/diodiogod/TTS-Audio-Suite.git
cd TTS-Audio-Suite
python install.py

No download - it's a tiny utility node, instant once the pack is present.

Common issues & troubleshooting

I added a voice file and it's still not in the dropdown. That's exactly the case this node fixes - but a caveat: ComfyUI dropdown widgets populate their options when the graph loads, so a brand-new voice may need a canvas refresh to appear as a selectable option too. This node refreshes the suite's internal voice cache so generation can find the file; if the dropdown itself is stale, reload the ComfyUI page. Between the two you avoid a full server restart.

The node didn't seem to run. ComfyUI skips nodes whose output nobody uses. If your Refresh Voice Cache node isn't firing, make sure its signal output is actually wired into something downstream that runs - a dangling refresh node gets pruned from execution.

It refreshed but generation still can't find the voice. Check the info output - it'll tell you what the refresh saw. If your new file isn't in the count, the problem is upstream: wrong folder, wrong format, or an incomplete save. The cache can only surface files that are actually there and valid.