Add realtime audio mapping support and SIP session payload tests

docs/llms-full.txt

@@ -33,7 +33,7 @@ The Agents SDK delivers a focused set of Python primitives—agents, tools, guar
 - [Voice quickstart](https://openai.github.io/openai-agents-python/voice/quickstart/): Build an end-to-end voice assistant with streaming transcription, text-to-speech, and event-driven responses.
 - [Voice pipeline](https://openai.github.io/openai-agents-python/voice/pipeline/): Customize audio capture, buffering, model invocation, and playback in voice-first experiences.
 - [Voice tracing](https://openai.github.io/openai-agents-python/voice/tracing/): Inspect voice session traces, latency breakdowns, and audio event timelines.
-- [Realtime quickstart](https://openai.github.io/openai-agents-python/realtime/quickstart/): Launch realtime agents over WebRTC or websockets, subscribe to events, and manage low-latency execution.
+- [Realtime quickstart](https://openai.github.io/openai-agents-python/realtime/quickstart/): Launch realtime agents over websockets (WebRTC is not available in the Python SDK), subscribe to events, and manage low-latency execution.
 - [Realtime guide](https://openai.github.io/openai-agents-python/realtime/guide/): Deep dive into realtime session lifecycle, event schemas, concurrency, and backpressure handling.
 
 ## Models and Provider Integrations

docs/llms.txt

@@ -36,7 +36,7 @@ The SDK focuses on a concise set of primitives so you can orchestrate multi-agen
 ## Modalities and Interfaces
 - [Voice quickstart](https://openai.github.io/openai-agents-python/voice/quickstart/): Build speech-enabled agents with streaming transcription and TTS.
 - [Voice pipeline](https://openai.github.io/openai-agents-python/voice/pipeline/): Customize audio ingestion, tool execution, and response rendering.
-- [Realtime quickstart](https://openai.github.io/openai-agents-python/realtime/quickstart/): Stand up low-latency realtime agents with WebRTC and websocket transports.
+- [Realtime quickstart](https://openai.github.io/openai-agents-python/realtime/quickstart/): Stand up low-latency realtime agents with websocket transport (WebRTC is not available in the Python SDK).
 - [Realtime guide](https://openai.github.io/openai-agents-python/realtime/guide/): Deep dive into session lifecycle, event formats, and concurrency patterns.
 
 ## API Reference Highlights

src/agents/realtime/__init__.py

@@ -84,6 +84,7 @@
 )
 from .openai_realtime import (
     DEFAULT_MODEL_SETTINGS,
+    OpenAIRealtimeSIPModel,
     OpenAIRealtimeWebSocketModel,
     get_api_key,
 )
@@ -176,6 +177,7 @@
     "RealtimeModelUserInputMessage",
     # OpenAI Realtime
     "DEFAULT_MODEL_SETTINGS",
+    "OpenAIRealtimeSIPModel",
     "OpenAIRealtimeWebSocketModel",
     "get_api_key",
     # Session

src/agents/realtime/audio_formats.py

@@ -1,5 +1,8 @@
 from __future__ import annotations
 
+from collections.abc import Mapping
+from typing import Any, Literal
+
 from openai.types.realtime.realtime_audio_formats import (
     AudioPCM,
     AudioPCMA,
@@ -11,7 +14,7 @@
 
 
 def to_realtime_audio_format(
-    input_audio_format: str | RealtimeAudioFormats | None,
+    input_audio_format: str | RealtimeAudioFormats | Mapping[str, Any] | None,
 ) -> RealtimeAudioFormats | None:
     format: RealtimeAudioFormats | None = None
     if input_audio_format is not None:
@@ -24,6 +27,27 @@ def to_realtime_audio_format(
                 format = AudioPCMA(type="audio/pcma")
             else:
                 logger.debug(f"Unknown input_audio_format: {input_audio_format}")
+        elif isinstance(input_audio_format, Mapping):
+            fmt_type = input_audio_format.get("type")
+            rate = input_audio_format.get("rate")
+            if fmt_type == "audio/pcm":
+                pcm_rate: Literal[24000] | None
+                if isinstance(rate, (int, float)) and int(rate) == 24000:
+                    pcm_rate = 24000
+                elif rate is None:
+                    pcm_rate = 24000
+                else:
+                    logger.debug(
+                        f"Unknown pcm rate in input_audio_format mapping: {input_audio_format}"
+                    )
+                    pcm_rate = 24000
+                format = AudioPCM(type="audio/pcm", rate=pcm_rate)
+            elif fmt_type == "audio/pcmu":
+                format = AudioPCMU(type="audio/pcmu")
+            elif fmt_type == "audio/pcma":
+                format = AudioPCMA(type="audio/pcma")
+            else:
+                logger.debug(f"Unknown input_audio_format mapping: {input_audio_format}")
         else:
             format = input_audio_format
     return format

src/agents/realtime/config.py

@@ -1,10 +1,7 @@
 from __future__ import annotations
 
-from typing import (
-    Any,
-    Literal,
-    Union,
-)
+from collections.abc import Mapping
+from typing import Any, Literal, Union
 
 from openai.types.realtime.realtime_audio_formats import (
     RealtimeAudioFormats as OpenAIRealtimeAudioFormats,
@@ -28,13 +25,20 @@
         "gpt-4o-realtime-preview-2024-12-17",
         "gpt-4o-realtime-preview-2024-10-01",
         "gpt-4o-mini-realtime-preview-2024-12-17",
+        "gpt-realtime-mini",
+        "gpt-realtime-mini-2025-10-06",
     ],
     str,
 ]
 """The name of a realtime model."""
 
 
-RealtimeAudioFormat: TypeAlias = Union[Literal["pcm16", "g711_ulaw", "g711_alaw"], str]
+RealtimeAudioFormat: TypeAlias = Union[
+    Literal["pcm16", "g711_ulaw", "g711_alaw"],
+    str,
+    Mapping[str, Any],
+    OpenAIRealtimeAudioFormats,
+]
 """The audio format for realtime audio streams."""
 
 
@@ -96,6 +100,30 @@ class RealtimeTurnDetectionConfig(TypedDict):
     """Threshold for server-vad to trigger a response if the user is idle for this duration."""
 
 
+class RealtimeAudioInputConfig(TypedDict, total=False):
+    """Configuration for audio input in realtime sessions."""
+
+    format: RealtimeAudioFormat | OpenAIRealtimeAudioFormats
+    noise_reduction: RealtimeInputAudioNoiseReductionConfig | None
+    transcription: RealtimeInputAudioTranscriptionConfig
+    turn_detection: RealtimeTurnDetectionConfig
+
+
+class RealtimeAudioOutputConfig(TypedDict, total=False):
+    """Configuration for audio output in realtime sessions."""
+
+    format: RealtimeAudioFormat | OpenAIRealtimeAudioFormats
+    voice: str
+    speed: float
+
+
+class RealtimeAudioConfig(TypedDict, total=False):
+    """Audio configuration for realtime sessions."""
+
+    input: RealtimeAudioInputConfig
+    output: RealtimeAudioOutputConfig
+
+
 class RealtimeSessionModelSettings(TypedDict):
     """Model settings for a realtime model session."""
 
@@ -111,6 +139,12 @@ class RealtimeSessionModelSettings(TypedDict):
     modalities: NotRequired[list[Literal["text", "audio"]]]
     """The modalities the model should support."""
 
+    output_modalities: NotRequired[list[Literal["text", "audio"]]]
+    """The output modalities the model should support."""
+
+    audio: NotRequired[RealtimeAudioConfig]
+    """The audio configuration for the session."""
+
     voice: NotRequired[str]
     """The voice to use for audio output."""