OpenAI Speech Generation

This reference graph uses generic Subfork primitives to call OpenAI text-to-speech and return previewable audio media.

Before running it, add a graph-scoped secret named OPENAI_API_KEY in Graph Settings > Execution Profile.

The graph shape is:

Text Value -> JSON Template -> HTTP Request -> Media Response

HTTP Request now exposes media for audio/*, image/*, and video/* responses by wrapping bounded response bytes as a data: URL. That keeps this graph usable without a dedicated OpenAI node or artifact store.

Expected outputs:

• audio points at the speech request node’s media output for Audio Preview panels.
• response points at Audio Response for API-style graph execution.

Known v0 caveat: inline data: URLs are fine for short local audio previews, but artifact storage should eventually replace large binary payloads.