[VoiceLive] Fix issue in function calling API (#42910)

* Fix issues in function calling

* update changelog

* update api docs for ResponseResource.create

* update Keyword-only separator for azure.ai.voicelive.ResponseResource.create

* update change log

* update

* update version

* update label

* fix too long line pylint error

* update toml

* ensure that we clean up after each venv runs

---------

Co-authored-by: Xiting Zhang <[email protected]>
Co-authored-by: Scott Beddall <[email protected]>

Author: 3 people

.github/CODEOWNERS

@@ -259,8 +259,8 @@
 # ServiceLabel: %AI Projects %Service Attention
 /sdk/ai/azure-ai-projects/                                           @dargilco @jhakulin @trangevi @glharper @nick863 @howieleung
 
-# PRLabel: %AI Voice Live
-# ServiceLabel: %AI Voice Live %Service Attention
+# PRLabel: %Voice Live
+# ServiceLabel: %Voice Live %Service Attention
 /sdk/ai/azure-ai-voicelive/                                          @rhurey @xitzhang
 
 

sdk/ai/azure-ai-voicelive/CHANGELOG.md

@@ -1,5 +1,14 @@
 # Release History
 
+## 1.0.0b2 (Unreleased)
+
+### Features Added
+
+- Async function call
+
+### Bugs Fixed
+- Fixed function calling: ensure `FunctionCallOutputItem.output` is properly serialized as a JSON string before sending to the service.
+
 ## 1.0.0b1 (2025-08-28)
 
 ### Features Added

sdk/ai/azure-ai-voicelive/apiview-properties.json

@@ -91,6 +91,8 @@
         "azure.ai.voicelive.models.ServerEventResponseContentPartDone": "VoiceLive.ServerEventResponseContentPartDone",
         "azure.ai.voicelive.models.ServerEventResponseCreated": "VoiceLive.ServerEventResponseCreated",
         "azure.ai.voicelive.models.ServerEventResponseDone": "VoiceLive.ServerEventResponseDone",
+        "azure.ai.voicelive.models.ServerEventResponseFunctionCallArgumentsDelta": "VoiceLive.ServerEventResponseFunctionCallArgumentsDelta",
+        "azure.ai.voicelive.models.ServerEventResponseFunctionCallArgumentsDone": "VoiceLive.ServerEventResponseFunctionCallArgumentsDone",
         "azure.ai.voicelive.models.ServerEventResponseOutputItemAdded": "VoiceLive.ServerEventResponseOutputItemAdded",
         "azure.ai.voicelive.models.ServerEventResponseOutputItemDone": "VoiceLive.ServerEventResponseOutputItemDone",
         "azure.ai.voicelive.models.ServerEventResponseTextDelta": "VoiceLive.ServerEventResponseTextDelta",

sdk/ai/azure-ai-voicelive/azure/ai/voicelive/_patch.py

@@ -31,6 +31,8 @@
     ClientEventResponseCancel,
     ClientEventResponseCreate,
     ClientEventSessionUpdate,
+    ConversationRequestItem,
+    ResponseCreateParams,
 )
 from azure.core.credentials import AzureKeyCredential, TokenCredential
 from azure.core.exceptions import AzureError
@@ -198,21 +200,33 @@ def __init__(self, connection: "VoiceLiveConnection") -> None:
         """
         self._connection = connection
 
-    def create(self, *, response: Optional[Mapping[str, Any]] = None, event_id: Optional[str] = None) -> None:
-        """Create a new response.
+    def create(
+        self,
+        *,
+        response: Optional[Union[ResponseCreateParams, Mapping[str, Any]]] = None,
+        event_id: Optional[str] = None,
+        additional_instructions: Optional[str] = None,
+    ) -> None:
+        """
+        Create a new response.
 
-        :keyword response: Optional mapping of response parameters to send.
-        :keyword type response: Mapping[str, Any] or None
+        :keyword response: Optional response parameters to send.
+        :keyword type response: ~azure.ai.voicelive.models.ResponseCreateParams or Mapping[str, Any] or None
         :keyword event_id: Optional ID for the event.
         :keyword type event_id: str or None
-        :return: None
+        :keyword additional_instructions: Extra system prompt appended to the session's default, for this response only.
+        :keyword type additional_instructions: str or None
         :rtype: None
         """
-        event = ClientEventResponseCreate()
-        if response:
-            event["response"] = dict(response)
-        if event_id:
-            event["event_id"] = event_id
+        if response is not None and not isinstance(response, ResponseCreateParams):
+            response = ResponseCreateParams(**dict(response))
+
+        # Use constructor kwargs (typed) instead of item-assignment
+        event = ClientEventResponseCreate(
+            event_id=event_id,
+            response=response,
+            additional_instructions=additional_instructions,
+        )
         self._connection.send(event)
 
     def cancel(self, *, response_id: Optional[str] = None, event_id: Optional[str] = None) -> None:
@@ -323,24 +337,31 @@ def __init__(self, connection: "VoiceLiveConnection") -> None:
         self._connection = connection
 
     def create(
-        self, *, item: Mapping[str, Any], previous_item_id: Optional[str] = None, event_id: Optional[str] = None
+        self,
+        *,
+        item: Mapping[str, Any],
+        previous_item_id: Optional[str] = None,
+        event_id: Optional[str] = None
     ) -> None:
         """Create a new conversation item.
 
         :keyword item: The item to create (message, function call, etc.).
-        :keyword type item: Mapping[str, Any]
+        :keyword ConversationRequestItem | Mapping[str, Any] item: The item to create (message/functions/etc.).
         :keyword previous_item_id: Insert after this item, if provided.
         :keyword type previous_item_id: str or None
         :keyword event_id: Optional ID for the event.
         :keyword type event_id: str or None
         :return: None
         :rtype: None
         """
-        event = ClientEventConversationItemCreate({"item": dict(item)})
-        if previous_item_id:
-            event["previous_item_id"] = previous_item_id
-        if event_id:
-            event["event_id"] = event_id
+        if not isinstance(item, ConversationRequestItem):
+            item = ConversationRequestItem(**dict(item))
+
+        event = ClientEventConversationItemCreate(
+            event_id=event_id,
+            previous_item_id=previous_item_id,
+            item=item,
+        )
         self._connection.send(event)
 
     def delete(self, *, item_id: str, event_id: Optional[str] = None) -> None:

sdk/ai/azure-ai-voicelive/azure/ai/voicelive/_version.py

@@ -6,4 +6,4 @@
 # Changes may cause incorrect behavior and will be lost if the code is regenerated.
 # --------------------------------------------------------------------------
 
-VERSION = "1.0.0b1"
+VERSION = "1.0.0b2"

sdk/ai/azure-ai-voicelive/azure/ai/voicelive/aio/_patch.py

@@ -33,6 +33,8 @@
     ClientEventResponseCancel,
     ClientEventResponseCreate,
     ClientEventSessionUpdate,
+    ConversationRequestItem,
+    ResponseCreateParams,
 )
 from azure.core.credentials import AzureKeyCredential, TokenCredential
 from azure.core.pipeline import policies
@@ -127,21 +129,34 @@ def __init__(self, connection: "VoiceLiveConnection") -> None:
         """
         self._connection = connection
 
-    async def create(self, *, response: Optional[Mapping[str, Any]] = None, event_id: Optional[str] = None) -> None:
+    async def create(
+        self,
+        *,
+        response: Optional[Union[ResponseCreateParams, Mapping[str, Any]]] = None,
+        event_id: Optional[str] = None,
+        additional_instructions: Optional[str] = None,
+    ) -> None:
         """Create a response from the model.
 
         This event instructs the server to create a Response (triggering model inference).
         When in Server VAD mode, the server may create responses automatically.
 
-        :keyword Mapping[str, Any] response: Optional response configuration.
-        :keyword str event_id: Optional ID for the event.
+        :keyword response: Optional response configuration to send.
+        :keyword type response: ~azure.ai.voicelive.models.ResponseCreateParams or Mapping[str, Any] or None
+        :keyword event_id: Optional ID for the event.
+        :keyword type event_id: str or None
+        :keyword additional_instructions: Extra system prompt appended to the session's default, for this response only.
+        :keyword type additional_instructions: str or None
         :rtype: None
         """
-        event = ClientEventResponseCreate()
-        if response is not None:
-            event["response"] = dict(response)
-        if event_id:
-            event["event_id"] = event_id
+        if response is not None and not isinstance(response, ResponseCreateParams):
+            response = ResponseCreateParams(**dict(response))
+
+        event = ClientEventResponseCreate(
+            event_id=event_id,
+            response=response,
+            additional_instructions=additional_instructions,
+        )
 
         await self._connection.send(event)
 
@@ -247,20 +262,27 @@ def __init__(self, connection: "VoiceLiveConnection") -> None:
         self._connection = connection
 
     async def create(
-        self, *, item: Mapping[str, Any], previous_item_id: Optional[str] = None, event_id: Optional[str] = None
+        self,
+        *,
+        item: Union[ConversationRequestItem, Mapping[str, Any]],
+        previous_item_id: Optional[str] = None,
+        event_id: Optional[str] = None
     ) -> None:
         """Create a new conversation item.
 
-        :keyword Mapping[str, Any] item: The item to create (message/functions/etc.).
+        :keyword ConversationRequestItem | Mapping[str, Any] item: The item to create (message/functions/etc.).
         :keyword str previous_item_id: Optional ID of the item after which to insert this item.
         :keyword str event_id: Optional ID for the event.
         :rtype: None
         """
-        event = ClientEventConversationItemCreate({"item": dict(item)})
-        if previous_item_id:
-            event["previous_item_id"] = previous_item_id
-        if event_id:
-            event["event_id"] = event_id
+        if not isinstance(item, ConversationRequestItem):
+            item = ConversationRequestItem(**dict(item))
+
+        event = ClientEventConversationItemCreate(
+            event_id=event_id,
+            previous_item_id=previous_item_id,
+            item=item,
+        )
         await self._connection.send(event)
 
     async def delete(self, *, item_id: str, event_id: Optional[str] = None) -> None:

sdk/ai/azure-ai-voicelive/azure/ai/voicelive/models/__init__.py

@@ -95,6 +95,8 @@
     ServerEventResponseContentPartDone,
     ServerEventResponseCreated,
     ServerEventResponseDone,
+    ServerEventResponseFunctionCallArgumentsDelta,
+    ServerEventResponseFunctionCallArgumentsDone,
     ServerEventResponseOutputItemAdded,
     ServerEventResponseOutputItemDone,
     ServerEventResponseTextDelta,
@@ -225,6 +227,8 @@
     "ServerEventResponseContentPartDone",
     "ServerEventResponseCreated",
     "ServerEventResponseDone",
+    "ServerEventResponseFunctionCallArgumentsDelta",
+    "ServerEventResponseFunctionCallArgumentsDone",
     "ServerEventResponseOutputItemAdded",
     "ServerEventResponseOutputItemDone",
     "ServerEventResponseTextDelta",

sdk/ai/azure-ai-voicelive/pyproject.toml

@@ -1,44 +1,34 @@
-# --------------------------------------------------------------------------
-# Copyright (c) Microsoft Corporation.
-# Licensed under the MIT License.
-# Code generated by Microsoft (R) Python Code Generator.
-# Changes may cause incorrect behavior and will be lost if the code is regenerated.
-# --------------------------------------------------------------------------
-
 [build-system]
 requires = ["setuptools>=77.0.3", "wheel"]
 build-backend = "setuptools.build_meta"
 
 [project]
 name = "azure-ai-voicelive"
-authors = [
-  { name = "Microsoft Corporation", email = "[email protected]" },
-]
-description = "Microsoft Corporation Azure Ai Voicelive Client Library for Python"
+authors = [{ name = "Microsoft Corporation", email = "[email protected]" }]
+description = "Microsoft Corporation Azure AI VoiceLive Client Library for Python"
 license = "MIT"
 classifiers = [
-    "Development Status :: 4 - Beta",
-    "Programming Language :: Python",
-    "Programming Language :: Python :: 3 :: Only",
-    "Programming Language :: Python :: 3",
-    "Programming Language :: Python :: 3.9",
-    "Programming Language :: Python :: 3.10",
-    "Programming Language :: Python :: 3.11",
-    "Programming Language :: Python :: 3.12",
-    "Programming Language :: Python :: 3.13",
+  "Development Status :: 4 - Beta",
+  "Programming Language :: Python",
+  "Programming Language :: Python :: 3 :: Only",
+  "Programming Language :: Python :: 3",
+  "Programming Language :: Python :: 3.9",
+  "Programming Language :: Python :: 3.10",
+  "Programming Language :: Python :: 3.11",
+  "Programming Language :: Python :: 3.12",
+  "Programming Language :: Python :: 3.13",
 ]
 requires-python = ">=3.9"
-keywords = ["azure", "azure sdk"]
+keywords = ["azure", "azure sdk", "voice", "voicelive", "realtime", "websocket", "audio"]
 
 dependencies = [
-    "isodate>=0.6.1",
-    "azure-core>=1.35.0",
-    "typing-extensions>=4.6.0",
-]
-dynamic = [
-"version", "readme"
+  "isodate>=0.6.1",
+  "azure-core>=1.35.1",                       # bump if you rely on 1.35.1
+  "typing-extensions>=4.6.0; python_version < '3.11'",
 ]
 
+dynamic = ["version", "readme"]
+
 [project.optional-dependencies]
 aiohttp = ["aiohttp>=3.9.0,<4.0.0"]
 websockets = ["websockets>=12.0,<14.0"]
@@ -57,15 +47,14 @@ test = [
   "azure-identity",
   "soundfile",
   "pyaudio; platform_python_implementation == 'CPython' and python_version < '3.13'",
-  # add "isodate" here only if you need a different test pin than runtime
 ]
 
 [project.urls]
 Repository = "https://github.com/Azure/azure-sdk-for-python"
 
 [tool.setuptools.dynamic]
 version = { attr = "azure.ai.voicelive._version.VERSION" }
-readme = {file = ["README.md", "CHANGELOG.md"], content-type = "text/markdown"}
+readme = { file = ["README.md", "CHANGELOG.md"], content-type = "text/markdown" }
 
 [tool.setuptools.packages.find]
 include = ["azure.ai.voicelive", "azure.ai.voicelive.*"]