OpenAI Speech Generation
This reference graph uses generic Subfork primitives to call OpenAI text-to-speech
and return previewable audio. Add a graph-scoped OPENAI_API_KEY secret before
running it.
The HTTP Request node explicitly sets response_storage to artifact, so a
successful audio response is stored as a temporary execution artifact even if its
artifact output is not connected to another node. The graph exposes that artifact
as asset. To keep it as an account asset beyond execution-artifact expiry,
retain the generated output from the graph’s Assets panel. Retained outputs count
against account storage quotas.
Provider calls may cost money. Re-running the graph creates a new paid speech request and a new temporary artifact.