ElevenLabs
Text-to-speech and transcription with timestamps and speaker diarization.
Use ElevenLabs voice models through DeepintShield for speech and transcription. This is an audio-only provider.
What works
- Speech (TTS) with streaming, and speech with timestamps (non-streaming)
- Transcription (STT, non-streaming) with diarization and word or character timestamps
- Voice settings, pronunciation dictionaries and output formats such as SRT, DOCX, PDF and segmented JSON
- List models
Limits to know: No chat, Responses, text completions, embeddings or image generation. A voice ID is required for TTS.
How to connect
Add your ElevenLabs API key in Model Hub → AI Providers, create a Virtual Key, then send speech or transcription requests with elevenlabs/<model>.
Read the ElevenLabs provider docs. Requests get the same Virtual Key budgets, guardrails and logs as every other provider on the AI gateway.