Realtime speech (WebSocket streaming)

Words as they are spoken, audio as it is generated — three WebSockets (duplex dialogue / realtime transcription / synthesis) plus `stream: true` on `/v1/audio/speech`. Events follow the OpenAI Realtime shape; billing is per segment, one usage row each.

• Raw markdown of this page: /docs/en/realtime-speech.md
• Docs index: /llms.txt

Machine-readable: /llms.txt, /models.md (append .md to any docs URL for raw markdown)