{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "$id": "https://jambonz.org/schema/verbs/voicelive_s2s",
  "minVersion": "11.1.0",
  "title": "Azure Voice Live S2S",
  "description": "Shortcut for 'llm' with vendor automatically set to 'voicelive'. Connects the caller to the Azure Voice Live API for real-time speech-to-speech conversation. connectOptions.host is required and names your Microsoft Foundry or Azure Speech resource (e.g. my-resource.services.ai.azure.com). auth takes either apiKey (sent as the api-key query parameter) or accessToken (a Microsoft Entra ID token, sent as a Bearer header). llmOptions.session_update is required and is sent verbatim in Voice Live's flat session shape, so Azure-only settings such as azure_semantic_vad, azure_deep_noise_suppression and Azure TTS voices are supported.",
  "type": "object",
  "allOf": [
    {
      "$ref": "../components/llm-base"
    }
  ],
  "properties": {
    "verb": {
      "const": "voicelive_s2s",
      "description": "The verb name."
    },
    "vendor": {
      "type": "string",
      "const": "voicelive",
      "description": "The LLM vendor (always 'voicelive' for this shortcut)."
    },
    "connectOptions": {
      "type": "object",
      "description": "Connection options for the Voice Live endpoint.",
      "properties": {
        "host": {
          "type": "string",
          "description": "Required. The Foundry/Speech resource hostname, e.g. my-resource.services.ai.azure.com."
        },
        "apiVersion": {
          "type": "string",
          "description": "The Voice Live api-version. Defaults to 2026-04-10.",
          "examples": [
            "2026-04-10",
            "2026-07-15"
          ]
        },
        "agentId": {
          "type": "string",
          "description": "Foundry Agent Service agent id. When set, the session is addressed by agent instead of model."
        },
        "projectId": {
          "type": "string",
          "description": "Foundry project id, used together with agentId."
        },
        "path": {
          "type": "string",
          "description": "Overrides the entire websocket path, including the query string. Rarely needed."
        }
      },
      "required": [
        "host"
      ],
      "additionalProperties": true
    },
    "auth": {
      "type": "object",
      "description": "Either apiKey (sent as the api-key query parameter) or accessToken (a Microsoft Entra ID token sent as a Bearer header).",
      "properties": {
        "apiKey": {
          "type": "string",
          "minLength": 1,
          "description": "Azure Voice Live API key."
        },
        "accessToken": {
          "type": "string",
          "minLength": 1,
          "description": "Microsoft Entra ID token for the https://ai.azure.com/.default scope."
        }
      },
      "anyOf": [
        {
          "required": [
            "apiKey"
          ]
        },
        {
          "required": [
            "accessToken"
          ]
        }
      ],
      "additionalProperties": true
    },
    "llmOptions": {
      "type": "object",
      "description": "session_update is sent verbatim in Voice Live's flat session shape; response_create triggers the first response.",
      "properties": {
        "session_update": {
          "type": "object",
          "description": "The flat Voice Live session body."
        },
        "response_create": {
          "type": "object",
          "description": "The initial response.create body."
        }
      },
      "required": [
        "session_update",
        "response_create"
      ],
      "additionalProperties": true
    }
  },
  "required": [
    "auth",
    "connectOptions",
    "llmOptions"
  ],
  "examples": [
    {
      "verb": "voicelive_s2s",
      "model": "gpt-realtime",
      "auth": {
        "apiKey": "..."
      },
      "connectOptions": {
        "host": "my-resource.services.ai.azure.com"
      },
      "llmOptions": {
        "session_update": {
          "modalities": [
            "text",
            "audio"
          ],
          "instructions": "You are a helpful voice assistant.",
          "voice": {
            "name": "en-US-Ava:DragonHDLatestNeural",
            "type": "azure-standard"
          },
          "turn_detection": {
            "type": "azure_semantic_vad",
            "silence_duration_ms": 500,
            "remove_filler_words": true
          },
          "input_audio_noise_reduction": {
            "type": "azure_deep_noise_suppression"
          }
        },
        "response_create": {}
      },
      "actionHook": "/s2s-complete"
    }
  ]
}