VoiceThere
Sign up

Azure Speech on VoiceThere

Azure conversation REST STT (finals only) and neural REST TTS on VoiceThere cloud with Azure Speech secrets.

Overview of all vendors: STT & TTS vendors.

What it is

Azure AI Speech provides REST conversation transcription and neural SSML synthesis. VoiceThere uses regional REST endpoints, not the Speech SDK or Voice Live realtime APIs.

Home: Azure Speech service

Set up on VoiceThere

  • Dashboard: Project Settings → Voice (STT / TTS). Pick azure on the leg this vendor supports. Add project secrets from the list below, then Deploy to cloud.
  • Secrets: Store keys as AGENT_* project secrets (see Agent environment & secrets).
  • API: Patch stt_provider / tts_provider on the project before deploy. See Control plane API.

Project secrets for this vendor

  • AGENT_AZURE_SPEECH_KEY: Azure Speech key. Subscription key (Ocp-Apim-Subscription-Key). You may also set AGENT_SPEECH_KEY → SPEECH_KEY.
  • AGENT_AZURE_SPEECH_REGION: Azure Speech region. Regional host segment (e.g. westeurope).
  • AGENT_AZURE_SPEECH_RESOURCE: Azure STT resource hostname. Cognitive Services hostname for REST STT (e.g. myresource.cognitiveservices.azure.com).

Optional env fields

  • AGENT_AZURE_SPEECH_LANGUAGE: STT language. BCP-47 language for Azure REST STT (default en-US).
  • AGENT_SPEECH_KEY: Speech key alias. Optional alias for AZURE_SPEECH_KEY (Ocp-Apim-Subscription-Key) when you prefer SPEECH_KEY on the runner.

CLI (after secrets exist):

voicethere projects voice set --project YOUR_PROJECT_ID \
  --stt-provider azure --tts-provider azure \
  --stt-model conversation --tts-voice en-US-JennyNeural

Models and voices you can pick

Project Voice settings in the dashboard lists every model and voice id VoiceThere accepts for this vendor. The same allowlists ship in the npm package @node-webrtc-rust/voice-catalog (pinned on your platform release).

STT examples: conversation

TTS voice examples: en-US-JennyNeural, af-ZA-AdriNeural

What the implementation does

STT posts utterance audio to the conversation REST API and reads final display text only (no streaming partials, no Speech SDK WebSocket frames, no Voice Live). TTS posts SSML to the neural REST synthesis endpoint for the voice ShortName you select.

Limitations

  • No Azure STT partials on VoiceThere. Microsoft documents SDK and Voice Live paths separately; this integration does not implement them.
  • Requires AGENT_AZURE_SPEECH_KEY, AGENT_AZURE_SPEECH_REGION, and AGENT_AZURE_SPEECH_RESOURCE (STT hostname). Optional AGENT_AZURE_SPEECH_LANGUAGE and AGENT_SPEECH_KEY alias.
  • Billed by Azure.

Official docs

← All documentation