Expanded cloud STT and TTS vendors
September 28, 2026 · release
VoiceThere projects can pair more cloud speech vendors per leg: Deepgram for both listen and Aura/Flux speak, ElevenLabs Scribe plus TTS, Groq Whisper and Orpheus, Azure Speech REST, and AWS Transcribe with Polly, in addition to OpenAI, Google, AssemblyAI, Cartesia, and on-cluster Sherpa.
Cloud vendor audio is billed by the provider you choose. Sherpa on the runner still uses VoiceThere usage credits.
Try it
- Docs hub: STT & TTS vendors (streaming matrix and links to each vendor guide)
- OpenAI: OpenAI speech on VoiceThere
- Deepgram: Deepgram speech on VoiceThere
- Groq: Groq speech on VoiceThere
- Azure: Azure Speech on VoiceThere
- Sherpa on-cluster: Sherpa on-cluster voice
- Dashboard: VoiceThere dashboard → your project → Voice (STT / TTS) → Deploy to cloud after changes
- CLI: fetch voice settings, then patch providers
voicethere projects voice show --project YOUR_PROJECT_ID
voicethere projects voice set --project YOUR_PROJECT_ID --stt-provider deepgram --tts-provider deepgram
After you change providers or secrets, redeploy the project so new sessions pick up the voice configuration.