Short answer: No. The ElevenLabs docs do not document a way to mint a temporary/ephemeral token specifically for the Text-to-Speech WebSocket. The TTS WebSocket requires an xi-api-key header; while it lists an optional authorization query parameter, no token-issuance method for TTS is provided. (elevenlabs.io)
What does exist today:
Single‑use (15‑minute) tokens are documented for Realtime Speech‑to‑Text and can be passed via the token query parameter on the STT WebSocket. (elevenlabs.io)
The Agents platform supports signed WebSocket URLs that embed a temporary token (expires after 15 minutes), but this applies to agent conversations, not TTS. (elevenlabs.io)
References:
TTS WebSocket API reference (auth via xi-api-key; optional authorization param, no TTS token minting method). (elevenlabs.io)
Single‑use token endpoint (expires after 15 minutes; currently documented type: realtime_scribe). (elevenlabs.io)
Realtime STT WebSocket (supports token from the single‑use token endpoint). (elevenlabs.io)
Agents authentication (signed URLs with 15‑minute temporary tokens). (elevenlabs.io)
barren thorn
past skiff