AllTalk Settings Loader

One of the silent traps in any "TTS via a separate server" setup is that the node you call doesn't tell you what the server is doing behind the scenes. AllTalk Settings Loader is the tiny node that fixes that: give it an alltalk_server_url and it GETs /api/currentsettings, then hands you a formatted string describing the server's actual state.

The output - a single settings_info STRING - reads like this, straight from the server response:

AllTalk Settings:
Engine: ...loaded
Model: ...loaded
DeepSpeed: False
LowVRAM: False
Streaming: ...
Multi-voice: ...

That's the whole node. One input, one output, no models, no secrets, no real failure modes beyond an unreachable server (which returns an error string rather than a crash).

When it earns its keep

TTS nodes inherit whatever the server has loaded, and "why does my voice sound wrong all of a sudden" is almost always answered by "because the server is running a different model/engine than last week." This node is how you check that without leaving the graph - especially if you run multiple AllTalk instances or share one with other tools. It's also a genuinely handy onboarding probe: first time you wire up AllTalk, drop this next to a Show Text node, run it once, and confirm the server actually came up before you start blaming the TTS generator.

It's a debug tool, not a workflow centerpiece. If you've only got one AllTalk server and it never changes, you don't need it. If you've ever had to wonder which engine was loaded, you'll be glad it's there.