AllTalk Voice Loader

This is a two-input, one-output helper that exists because of a mismatch in the pack's own design. The AllTalk TTS Generator builds its voice dropdown from a local folder on your machine, but the voices that actually work are the ones the AllTalk server has loaded. AllTalk Voice Loader asks the server directly and tells you the truth.

It's about as simple as a node gets: point alltalk_server_url at your server (default http://localhost:7851), and it GETs /api/voices. On success you get a voices_info STRING formatted like:

Available voices:
female_01.wav
male_01.wav
...

If the request fails, you get an error string instead of a crash - "Error connecting to AllTalk" when the server is unreachable, or the server's own message when it is. Note this node is not marked as an output node, so nothing pops in a UI panel until you wire it somewhere - feed it into a Show Text-style node to read the list.

Why you'd bother

Two honest reasons. First, debugging: the voice you typed into the TTS generator isn't speaking, and you want to know whether it's a name mismatch or a server problem. Second, drift: AllTalk lets you add voices and swap models in its own interface, and the local voices_dir the generator reads can fall behind what the server can actually produce. This node is the server-side source of truth.

It also costs you nothing - no models, no API keys, and it runs against the same AllTalk install the generator uses. If you're not running AllTalk at all, skip it; if you are, it's a one-click sanity check that pays for itself the first time a voice stops working.