ElevenLabs Refresh Voices

This is the maintenance node you'll reach for exactly when things look broken. The pack caches your voice list for an hour, and that cache is why a freshly cloned voice doesn't show up in the TTS dropdown, why switching accounts leaves stale voices behind, and why a brand-new install sometimes looks empty. Refresh Voices exists to punch that cache in the face on demand.

How it works

The pack stores the fetched voice and model lists in a shared in-process cache with a one-hour TTL. This node forces both caches invalid, then refetches from the API - refresh_voices and refresh_models are two independent toggles, each defaulting to on, so you can refresh just the list you care about. The result is a status STRING telling you how many voices and models were refreshed, echoed to the console.

There's a quirk worth knowing: the "models" list isn't fetched from any live endpoint - ElevenLabs has no models list API, so the pack ships a hardcoded list of known model IDs (eleven_v3, the turbo/flash family, etc.). Refreshing models basically re-sets that static list. It's harmless, but it means refresh_models isn't going to reveal brand-new models until the pack itself updates.

Inputs

• api_key - your key (used for the voice fetch).
• refresh_voices - boolean, default true.
• refresh_models - boolean, default true.

Output: a status STRING.

When you actually need it

Three cases, and all three are "you just changed something about your account":

1. You cloned a voice - run this, then reload the TTS node's dropdown. The clone appears.
2. You switched API keys or accounts - the TTS and Voice Changer nodes auto-refresh on key change, but if you're using a fresh key in a workflow that didn't trigger it, this is the manual override.
3. The dropdown is empty or wrong on a fresh install - voices don't populate until something fetches them; this forces it.

Install

Same pack, same command:

cd ComfyUI/custom_nodes
git clone https://github.com/karthikg-09/Elevenlabs-ComfyUI.git
cd ElevenLabs-ComfyUI && pip install -r requirements.txt

or ComfyUI Manager → "Elevenlabs-ComfyUI" → restart. Add Node → ElevenLabs.

One caveat about expectations: this node refreshes the cache, not the dropdown widgets already on your canvas. After it runs, you still need to right-click → Reload Node (or F5) on the TTS node to see the new list. Refresh the cache here, reload the node there, and the "empty dropdown" complaint dies for good.