sora-sdk 2024.4.0.dev15__tar.gz → 2024.4.0.dev18__tar.gz

{sora_sdk-2024.4.0.dev15/src/sora_sdk.egg-info → sora_sdk-2024.4.0.dev18}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sora_sdk
-Version: 2024.4.0.dev15
+Version: 2024.4.0.dev18
 Summary: WebRTC SFU Sora Python SDK
 Home-page: https://github.com/shiguredo/sora-python-sdk
 Author-email: "Shiguredo Inc." <contact+pypi@shiguredo.jp>
@@ -226,7 +226,8 @@ Please read <https://github.com/shiguredo/oss/blob/master/README.en.md> before u
 - [Sora C++ SDK](https://github.com/shiguredo/sora-cpp-sdk) ベース
 - WebRTC 部分の機能は [libwebrtc](https://webrtc.googlesource.com/src/) を採用
 - Windows / macOS / Linux (Ubuntu) プラットフォームに対応
-- WebRTC 統計情報の取得が可能
+- [WebRTC 統計情報](https://www.w3.org/TR/webrtc-stats/) の取得が可能
+- [WebRTC Encoded Transform](https://www.w3.org/TR/webrtc-encoded-transform/) に対応
 - Intel / Apple / NVIDIA のハードウェアデコーダー/エンコーダーに対応
   - Intel VPL (AV1 / H.264 / H.265)
   - Apple Video Toolbox (H.264 / H.265)
@@ -264,7 +265,6 @@ Please read <https://github.com/shiguredo/oss/blob/master/README.en.md> before u
 
 ```bash
 uv add sora_sdk
-uv sync
 ```
 
 ### pip
@@ -311,8 +311,6 @@ PyPI 経由ではインストールできません。
 
 ### 優先実装が可能な機能一覧
 
-**詳細は Discord やメールなどでお気軽にお問い合わせください**
-
 - Windows 11 arm64
 - Ubuntu 24.04 arm64
 - Ubuntu 22.04 arm64

{sora_sdk-2024.4.0.dev15 → sora_sdk-2024.4.0.dev18}/README.md

@@ -28,7 +28,8 @@ Please read <https://github.com/shiguredo/oss/blob/master/README.en.md> before u
 - [Sora C++ SDK](https://github.com/shiguredo/sora-cpp-sdk) ベース
 - WebRTC 部分の機能は [libwebrtc](https://webrtc.googlesource.com/src/) を採用
 - Windows / macOS / Linux (Ubuntu) プラットフォームに対応
-- WebRTC 統計情報の取得が可能
+- [WebRTC 統計情報](https://www.w3.org/TR/webrtc-stats/) の取得が可能
+- [WebRTC Encoded Transform](https://www.w3.org/TR/webrtc-encoded-transform/) に対応
 - Intel / Apple / NVIDIA のハードウェアデコーダー/エンコーダーに対応
   - Intel VPL (AV1 / H.264 / H.265)
   - Apple Video Toolbox (H.264 / H.265)
@@ -66,7 +67,6 @@ Please read <https://github.com/shiguredo/oss/blob/master/README.en.md> before u
 
 ```bash
 uv add sora_sdk
-uv sync
 ```
 
 ### pip
@@ -113,8 +113,6 @@ PyPI 経由ではインストールできません。
 
 ### 優先実装が可能な機能一覧
 
-**詳細は Discord やメールなどでお気軽にお問い合わせください**
-
 - Windows 11 arm64
 - Ubuntu 24.04 arm64
 - Ubuntu 22.04 arm64

{sora_sdk-2024.4.0.dev15 → sora_sdk-2024.4.0.dev18}/pyproject.toml

@@ -1,7 +1,7 @@
 [project]
 name = "sora_sdk"
 authors = [{ name = "Shiguredo Inc.", email = "contact+pypi@shiguredo.jp" }]
-version = "2024.4.0.dev15"
+version = "2024.4.0.dev18"
 description = "WebRTC SFU Sora Python SDK"
 readme = "README.md"
 license = { file = "LICENSE" }

{sora_sdk-2024.4.0.dev15 → sora_sdk-2024.4.0.dev18}/src/sora_sdk/sora_sdk_ext.pyi

@@ -8,7 +8,7 @@ from numpy.typing import ArrayLike
 class Sora:
     def __init__(self, use_hardware_encoder: bool | None = None, openh264: str | None = None) -> None: ...
 
-    def create_connection(self, signaling_urls: list[str], role: str, channel_id: str, client_id: Optional[str] = None, bundle_id: Optional[str] = None, metadata: Optional[dict] = None, signaling_notify_metadata: Optional[dict] = None, audio_source: Optional[SoraTrackInterface] = None, video_source: Optional[SoraTrackInterface] = None, audio: Optional[bool] = None, video: Optional[bool] = None, audio_codec_type: Optional[str] = None, video_codec_type: Optional[str] = None, video_bit_rate: Optional[int] = None, audio_bit_rate: Optional[int] = None, video_vp9_params: Optional[dict] = None, video_av1_params: Optional[dict] = None, video_h264_params: Optional[dict] = None, simulcast: Optional[bool] = None, spotlight: Optional[bool] = None, spotlight_number: Optional[int] = None, simulcast_rid: Optional[str] = None, spotlight_focus_rid: Optional[str] = None, spotlight_unfocus_rid: Optional[str] = None, forwarding_filter: Optional[dict] = None, data_channels: Optional[list[dict]] = None, data_channel_signaling: Optional[bool] = None, ignore_disconnect_websocket: Optional[bool] = None, data_channel_signaling_timeout: Optional[int] = None, disconnect_wait_timeout: Optional[int] = None, websocket_close_timeout: Optional[int] = None, websocket_connection_timeout: Optional[int] = None, audio_streaming_language_code: Optional[str] = None, insecure: Optional[bool] = None, client_cert: Optional[bytes] = None, client_key: Optional[bytes] = None, ca_cert: Optional[bytes] = None, proxy_url: Optional[str] = None, proxy_username: Optional[str] = None, proxy_password: Optional[str] = None, proxy_agent: Optional[str] = None) -> SoraConnection: ...
+    def create_connection(self, signaling_urls: list[str], role: str, channel_id: str, client_id: Optional[str] = None, bundle_id: Optional[str] = None, metadata: Optional[dict] = None, signaling_notify_metadata: Optional[dict] = None, audio_source: Optional[SoraTrackInterface] = None, video_source: Optional[SoraTrackInterface] = None, audio_frame_transformer: Optional[SoraAudioFrameTransformer] = None, video_frame_transformer: Optional[SoraVideoFrameTransformer] = None, audio: Optional[bool] = None, video: Optional[bool] = None, audio_codec_type: Optional[str] = None, video_codec_type: Optional[str] = None, video_bit_rate: Optional[int] = None, audio_bit_rate: Optional[int] = None, video_vp9_params: Optional[dict] = None, video_av1_params: Optional[dict] = None, video_h264_params: Optional[dict] = None, simulcast: Optional[bool] = None, spotlight: Optional[bool] = None, spotlight_number: Optional[int] = None, simulcast_rid: Optional[str] = None, spotlight_focus_rid: Optional[str] = None, spotlight_unfocus_rid: Optional[str] = None, forwarding_filter: Optional[dict] = None, data_channels: Optional[list[dict]] = None, data_channel_signaling: Optional[bool] = None, ignore_disconnect_websocket: Optional[bool] = None, data_channel_signaling_timeout: Optional[int] = None, disconnect_wait_timeout: Optional[int] = None, websocket_close_timeout: Optional[int] = None, websocket_connection_timeout: Optional[int] = None, audio_streaming_language_code: Optional[str] = None, insecure: Optional[bool] = None, client_cert: Optional[bytes] = None, client_key: Optional[bytes] = None, ca_cert: Optional[bytes] = None, proxy_url: Optional[str] = None, proxy_username: Optional[str] = None, proxy_password: Optional[str] = None, proxy_agent: Optional[str] = None) -> SoraConnection: ...
 
     def create_audio_source(self, channels: int, sample_rate: int) -> SoraAudioSource: ...
 
@@ -33,6 +33,17 @@ class SoraAudioFrame:
 
     def data(self) -> Annotated[ArrayLike, dict(dtype='int16', shape=(None, None))]: ...
 
+class SoraAudioFrameTransformer(SoraFrameTransformer):
+    def __init__(self) -> None: ...
+
+    def __del__(self) -> None: ...
+
+    @property
+    def on_transform(self) -> Callable[[SoraTransformableAudioFrame], None]: ...
+
+    @on_transform.setter
+    def on_transform(self, arg: Callable[[SoraTransformableAudioFrame], None], /) -> None: ...
+
 class SoraAudioSinkImpl:
     def __init__(self, track: SoraTrackInterface, output_frequency: int = -1, output_channels: int = 0) -> None: ...
 
@@ -145,6 +156,11 @@ class SoraConnection:
     @on_data_channel.setter
     def on_data_channel(self, arg: Callable[[str], None], /) -> None: ...
 
+class SoraFrameTransformer:
+    def enqueue(self, arg: SoraTransformableFrame, /) -> None: ...
+
+    def start_short_circuiting(self) -> None: ...
+
 class SoraLoggingSeverity(enum.IntEnum):
     VERBOSE = 0
 
@@ -160,6 +176,8 @@ class SoraMediaTrack(SoraTrackInterface):
     @property
     def stream_id(self) -> str: ...
 
+    def set_frame_transformer(self, arg: SoraFrameTransformer, /) -> None: ...
+
 class SoraSignalingDirection(enum.IntEnum):
     SENT = 0
 
@@ -209,6 +227,87 @@ class SoraTrackState(enum.IntEnum):
 
     ENDED = 1
 
+class SoraTransformableAudioFrame(SoraTransformableFrame):
+    @property
+    def contributing_sources(self) -> Annotated[ArrayLike, dict(dtype='uint32', shape=(None), writable=False)]: ...
+
+    @property
+    def sequence_number(self) -> int | None: ...
+
+    @property
+    def absolute_capture_timestamp(self) -> int | None: ...
+
+    @property
+    def type(self) -> SoraTransformableAudioFrameType: ...
+
+    @property
+    def audio_level(self) -> int | None: ...
+
+    @property
+    def receive_time(self) -> int | None: ...
+
+class SoraTransformableAudioFrameType(enum.IntEnum):
+    EMPTY = 0
+
+    SPEECH = 1
+
+    CN = 2
+
+class SoraTransformableFrame:
+    def get_data(self) -> Annotated[ArrayLike, dict(dtype='uint8', shape=(None), writable=False)]: ...
+
+    def set_data(self, arg: Annotated[ArrayLike, dict(dtype='uint8', shape=(None), order='C', device='cpu', writable=False)], /) -> None: ...
+
+    @property
+    def payload_type(self) -> int: ...
+
+    @property
+    def ssrc(self) -> int: ...
+
+    @property
+    def rtp_timestamp(self) -> int: ...
+
+    @rtp_timestamp.setter
+    def rtp_timestamp(self, arg: int, /) -> None: ...
+
+    @property
+    def direction(self) -> SoraTransformableFrameDirection: ...
+
+    @property
+    def mine_type(self) -> str: ...
+
+class SoraTransformableFrameDirection(enum.IntEnum):
+    UNKNOWN = 0
+
+    RECEIVER = 1
+
+    SENDER = 2
+
+class SoraTransformableVideoFrame(SoraTransformableFrame):
+    @property
+    def is_key_frame(self) -> bool: ...
+
+    @property
+    def frame_id(self) -> int | None: ...
+
+    @property
+    def frame_dependencies(self) -> Annotated[ArrayLike, dict(dtype='int64', shape=(None), writable=False)]: ...
+
+    @property
+    def width(self) -> int: ...
+
+    @property
+    def height(self) -> int: ...
+
+    @property
+    def spatial_index(self) -> int: ...
+
+    @property
+    def temporal_index(self) -> int: ...
+
+    @property
+    def contributing_sources(self) -> Annotated[ArrayLike, dict(dtype='uint32', shape=(None), writable=False)]: ...
+
 class SoraVAD:
     def __init__(self) -> None: ...
 
@@ -217,6 +316,17 @@ class SoraVAD:
 class SoraVideoFrame:
     def data(self) -> Annotated[ArrayLike, dict(dtype='uint8', shape=(None, None, 3))]: ...
 
+class SoraVideoFrameTransformer(SoraFrameTransformer):
+    def __init__(self) -> None: ...
+
+    def __del__(self) -> None: ...
+
+    @property
+    def on_transform(self) -> Callable[[SoraTransformableVideoFrame], None]: ...
+
+    @on_transform.setter
+    def on_transform(self, arg: Callable[[SoraTransformableVideoFrame], None], /) -> None: ...
+
 class SoraVideoSinkImpl:
     def __init__(self, arg: SoraTrackInterface, /) -> None: ...
 

{sora_sdk-2024.4.0.dev15 → sora_sdk-2024.4.0.dev18/src/sora_sdk.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: sora_sdk
-Version: 2024.4.0.dev15
+Version: 2024.4.0.dev18
 Summary: WebRTC SFU Sora Python SDK
 Home-page: https://github.com/shiguredo/sora-python-sdk
 Author-email: "Shiguredo Inc." <contact+pypi@shiguredo.jp>
@@ -226,7 +226,8 @@ Please read <https://github.com/shiguredo/oss/blob/master/README.en.md> before u
 - [Sora C++ SDK](https://github.com/shiguredo/sora-cpp-sdk) ベース
 - WebRTC 部分の機能は [libwebrtc](https://webrtc.googlesource.com/src/) を採用
 - Windows / macOS / Linux (Ubuntu) プラットフォームに対応
-- WebRTC 統計情報の取得が可能
+- [WebRTC 統計情報](https://www.w3.org/TR/webrtc-stats/) の取得が可能
+- [WebRTC Encoded Transform](https://www.w3.org/TR/webrtc-encoded-transform/) に対応
 - Intel / Apple / NVIDIA のハードウェアデコーダー/エンコーダーに対応
   - Intel VPL (AV1 / H.264 / H.265)
   - Apple Video Toolbox (H.264 / H.265)
@@ -264,7 +265,6 @@ Please read <https://github.com/shiguredo/oss/blob/master/README.en.md> before u
 
 ```bash
 uv add sora_sdk
-uv sync
 ```
 
 ### pip
@@ -311,8 +311,6 @@ PyPI 経由ではインストールできません。
 
 ### 優先実装が可能な機能一覧
 
-**詳細は Discord やメールなどでお気軽にお問い合わせください**
-
 - Windows 11 arm64
 - Ubuntu 24.04 arm64
 - Ubuntu 22.04 arm64

{sora_sdk-2024.4.0.dev15 → sora_sdk-2024.4.0.dev18}/src/sora_sdk.egg-info/SOURCES.txt

@@ -16,6 +16,7 @@ src/sora_sdk.egg-info/SOURCES.txt
 src/sora_sdk.egg-info/dependency_links.txt
 src/sora_sdk.egg-info/top_level.txt
 tests/test_ca_cert.py
+tests/test_encoded_transform.py
 tests/test_macos.py
 tests/test_messaging.py
 tests/test_openh264.py

sora_sdk-2024.4.0.dev18/tests/test_encoded_transform.py

@@ -0,0 +1,438 @@
+import json
+import sys
+import threading
+import time
+import uuid
+from threading import Event
+from typing import Any, Optional
+
+import numpy
+
+from sora_sdk import (
+    Sora,
+    SoraAudioFrameTransformer,
+    SoraAudioSource,
+    SoraMediaTrack,
+    SoraTransformableAudioFrame,
+    SoraTransformableVideoFrame,
+    SoraVideoFrameTransformer,
+    SoraVideoSource,
+)
+
+
+class SendonlyEncodedTransform:
+    def __init__(
+        self,
+        signaling_urls: list[str],
+        channel_id: str,
+        metadata: dict[str, Any],
+    ):
+        self._signaling_urls: list[str] = signaling_urls
+        self._channel_id: str = channel_id
+
+        self._connection_id: str
+
+        # 接続した
+        self._connected: Event = Event()
+        # 終了
+        self._closed = Event()
+
+        self._audio_channels: int = 1
+        self._audio_sample_rate: int = 16000
+
+        self._video_width: int = 960
+        self._video_height: int = 540
+
+        self._sora = Sora()
+
+        self._fake_audio_thread: Optional[threading.Thread] = None
+        self._fake_video_thread: Optional[threading.Thread] = None
+
+        self._audio_source: Optional[SoraAudioSource] = None
+        self._audio_source = self._sora.create_audio_source(
+            self._audio_channels, self._audio_sample_rate
+        )
+
+        self._video_source: Optional[SoraVideoSource] = None
+        self._video_source = self._sora.create_video_source()
+
+        # Audio 向けの Encoded Transformer
+        self._audio_transformer = SoraAudioFrameTransformer()
+        # Audio のエンコードフレームを受け取るコールバック関数を on_transform に設定
+        self._audio_transformer.on_transform = self._on_audio_transform
+
+        # Video 向けの Encoded Transformer
+        self._video_transformer = SoraVideoFrameTransformer()
+        # Video のエンコードフレームを受け取るコールバック関数を on_transform に設定
+        self._video_transformer.on_transform = self._on_video_transform
+
+        self._connection = self._sora.create_connection(
+            signaling_urls=signaling_urls,
+            role="sendonly",
+            channel_id=channel_id,
+            metadata=metadata,
+            audio=True,
+            video=True,
+            audio_source=self._audio_source,
+            video_source=self._video_source,
+            audio_frame_transformer=self._audio_transformer,
+            video_frame_transformer=self._video_transformer,
+        )
+
+        self._connection.on_set_offer = self._on_set_offer
+        self._connection.on_notify = self._on_notify
+        self._connection.on_disconnect = self._on_disconnect
+
+        self._is_called_on_audio_transform = False
+        self._is_called_on_video_transform = False
+
+    def connect(self):
+        self._connection.connect()
+
+        self._fake_audio_thread = threading.Thread(target=self._fake_audio_loop, daemon=True)
+        self._fake_audio_thread.start()
+
+        self._fake_video_thread = threading.Thread(target=self._fake_video_loop, daemon=True)
+        self._fake_video_thread.start()
+
+        # _connected が set されるまで 30 秒待つ
+        assert self._connected.wait(30)
+
+        return self
+
+    def disconnect(self):
+        self._connection.disconnect()
+
+    def get_stats(self):
+        raw_stats = self._connection.get_stats()
+        stats = json.loads(raw_stats)
+        return stats
+
+    @property
+    def is_called_on_audio_transform(self):
+        return self._is_called_on_audio_transform
+
+    @property
+    def is_called_on_video_transform(self):
+        return self._is_called_on_video_transform
+
+    def _fake_audio_loop(self):
+        while not self._closed.is_set():
+            time.sleep(0.02)
+            if self._audio_source is not None:
+                self._audio_source.on_data(numpy.zeros((320, 1), dtype=numpy.int16))
+
+    def _fake_video_loop(self):
+        while not self._closed.is_set():
+            time.sleep(1.0 / 30)
+            if self._video_source is not None:
+                self._video_source.on_captured(
+                    numpy.zeros((self._video_height, self._video_width, 3), dtype=numpy.uint8)
+                )
+
+    def _on_set_offer(self, raw_offer):
+        offer = json.loads(raw_offer)
+        if offer["type"] == "offer":
+            self._connection_id = offer["connection_id"]
+            print(f"Received 'Offer': connection_id={self._connection_id}")
+
+    def _on_notify(self, raw_message):
+        message = json.loads(raw_message)
+        if (
+            message["type"] == "notify"
+            and message["event_type"] == "connection.created"
+            and message["connection_id"] == self._connection_id
+        ):
+            print(f"Connected Sora: connection_id={self._connection_id}")
+            self._connected.set()
+
+    def _on_disconnect(self, error_code, message):
+        print(f"Disconnected Sora: error_code='{error_code}' message='{message}'")
+        self._closed.set()
+        self._connected.clear()
+
+        if self._fake_audio_thread is not None:
+            self._fake_audio_thread.join(timeout=10)
+
+        if self._fake_video_thread is not None:
+            self._fake_video_thread.join(timeout=10)
+
+    def _on_audio_transform(self, frame: SoraTransformableAudioFrame):
+        # この実装が Encoded Transform を利用する上での基本形となる
+
+        # frame からエンコードされたフレームデータを取得する
+        # 戻り値は numpy.ndarray になっている
+        new_data = frame.get_data()
+
+        # "sora" という文字列を new_data の後ろに追加
+        new_data = numpy.append(new_data, numpy.frombuffer(b"sora", dtype=numpy.uint8))
+
+        self._is_called_on_audio_transform = True
+
+        # ここで new_data の末尾にデータをつける new_data を暗号化するなど任意の処理を実装する
+
+        # 加工したフレームデータで frame の フレームデータを入れ替える
+        frame.set_data(new_data)
+        self._audio_transformer.enqueue(frame)
+
+    def _on_video_transform(self, frame: SoraTransformableVideoFrame):
+        # この実装が Encoded Transform を利用する上での基本形となる
+
+        # frame からエンコードされたフレームデータを取得する
+        # 戻り値は numpy.ndarray になっている
+        new_data = frame.get_data()
+
+        # "sora" という文字列を new_data の後ろに追加
+        new_data = numpy.append(new_data, numpy.frombuffer(b"sora", dtype=numpy.uint8))
+
+        self._is_called_on_video_transform = True
+
+        # ここで new_data の末尾にデータをつける new_data を暗号化するなど任意の処理を実装する
+
+        # 加工したフレームデータで frame の フレームデータを入れ替える
+        frame.set_data(new_data)
+        self._video_transformer.enqueue(frame)
+
+
+class RecvonlyEncodedTransform:
+    def __init__(
+        self,
+        signaling_urls: list[str],
+        channel_id: str,
+        metadata: dict[str, Any],
+    ):
+        self._signaling_urls: list[str] = signaling_urls
+        self._channel_id: str = channel_id
+
+        self._connection_id: str
+
+        # 接続した
+        self._connected: Event = Event()
+        # 終了
+        self._closed = Event()
+
+        self._audio_output_frequency: int = 24000
+        self._audio_output_channels: int = 1
+
+        self._sora = Sora()
+
+        self._connection = self._sora.create_connection(
+            signaling_urls=signaling_urls,
+            role="recvonly",
+            channel_id=channel_id,
+            metadata=metadata,
+            audio=True,
+            video=True,
+        )
+
+        self._connection.on_set_offer = self._on_set_offer
+        self._connection.on_notify = self._on_notify
+        self._connection.on_disconnect = self._on_disconnect
+
+        self._connection.on_track = self._on_track
+
+        self._is_called_on_audio_transform = False
+        self._is_called_on_video_transform = False
+
+    def connect(self):
+        self._connection.connect()
+
+        # _connected が set されるまで 30 秒待つ
+        assert self._connected.wait(30)
+
+        return self
+
+    def disconnect(self):
+        self._connection.disconnect()
+
+    def get_stats(self):
+        raw_stats = self._connection.get_stats()
+        stats = json.loads(raw_stats)
+        return stats
+
+    @property
+    def is_called_on_audio_transform(self):
+        return self._is_called_on_audio_transform
+
+    @property
+    def is_called_on_video_transform(self):
+        return self._is_called_on_video_transform
+
+    def _on_set_offer(self, raw_offer):
+        offer = json.loads(raw_offer)
+        if offer["type"] == "offer":
+            self._connection_id = offer["connection_id"]
+            print(f"Received 'Offer': connection_id={self._connection_id}")
+
+    def _on_notify(self, raw_message):
+        message = json.loads(raw_message)
+        if (
+            message["type"] == "notify"
+            and message["event_type"] == "connection.created"
+            and message["connection_id"] == self._connection_id
+        ):
+            print(f"Connected Sora: connection_id={self._connection_id}")
+            self._connected.set()
+
+    def _on_disconnect(self, error_code, message):
+        print(f"Disconnected Sora: error_code='{error_code}' message='{message}'")
+        self._closed = True
+        self._connected.clear()
+
+    def _on_track(self, track: SoraMediaTrack) -> None:
+        if track.kind == "audio":
+            # Audio 向けの Encoded Transformer
+            self._audio_transformer = SoraAudioFrameTransformer()
+            # Audio のエンコードフレームを受け取るコールバック関数を on_transform に設定
+            self._audio_transformer.on_transform = self._on_audio_transform
+            # Encoded Transformer を RTPReceiver に設定する
+            track.set_frame_transformer(self._audio_transformer)
+        if track.kind == "video":
+            # Video 向けの Encoded Transformer
+            self._video_transformer = SoraVideoFrameTransformer()
+            # Video のエンコードフレームを受け取るコールバック関数を on_transform に設定
+            self._video_transformer.on_transform = self._on_video_transform
+            # Encoded Transformer を SoraMediaTrack に設定する
+            track.set_frame_transformer(self._video_transformer)
+
+    def _on_audio_transform(self, frame: SoraTransformableAudioFrame):
+        # この実装が Encoded Transform を利用する上での基本形となる
+
+        # frame からエンコードされたフレームデータを取得する
+        # 戻り値は ArrayLike になっている
+        new_data = frame.get_data()
+
+        # ここで new_data の末尾にデータをつける new_data を暗号化するなど任意の処理を実装する
+
+        # ArrayLike を numpy.uint8 のバイト列に変換する
+        new_data = numpy.asarray(new_data, dtype=numpy.uint8)
+
+        # 後ろ4バイトを取得する
+        removed_data = new_data[-4:]
+
+        assert b"sora" == removed_data.tobytes()
+
+        # 後ろ4バイトを取り除く
+        new_data = new_data[:-4]
+
+        self._is_called_on_audio_transform = True
+
+        # 加工したフレームデータで frame の フレームデータを入れ替える
+        frame.set_data(new_data)
+        self._audio_transformer.enqueue(frame)
+
+    def _on_video_transform(self, frame: SoraTransformableVideoFrame):
+        # この実装が Encoded Transform を利用する上での基本形となる
+        # frame からエンコードされたフレームデータを取得する
+        # 戻り値は ArrayLike になっている
+        new_data = frame.get_data()
+
+        # ここで new_data の末尾にデータをつける new_data を暗号化するなど任意の処理を実装する
+
+        # ArrayLike を numpy.uint8 のバイト列に変換する
+        new_data = numpy.asarray(new_data, dtype=numpy.uint8)
+
+        # 後ろ4バイトを取得する
+        removed_data = new_data[-4:]
+
+        assert b"sora" == removed_data.tobytes()
+
+        # 後ろ4バイトを取り除く
+        new_data = new_data[:-4]
+
+        self._is_called_on_video_transform = True
+
+        # 加工したフレームデータで frame の フレームデータを入れ替える
+        frame.set_data(new_data)
+        self._video_transformer.enqueue(frame)
+
+
+def test_encoded_transform(setup):
+    signaling_urls = setup.get("signaling_urls")
+    channel_id_prefix = setup.get("channel_id_prefix")
+    metadata = setup.get("metadata")
+
+    channel_id = f"{channel_id_prefix}_{__name__}_{sys._getframe().f_code.co_name}_{uuid.uuid4()}"
+
+    sendonly = SendonlyEncodedTransform(
+        signaling_urls,
+        channel_id,
+        metadata,
+    )
+    sendonly.connect()
+
+    recvonly = RecvonlyEncodedTransform(
+        signaling_urls,
+        channel_id,
+        metadata,
+    )
+    recvonly.connect()
+
+    time.sleep(5)
+
+    sendonly_stats = sendonly.get_stats()
+    recvonly_stats = recvonly.get_stats()
+
+    sendonly.disconnect()
+    recvonly.disconnect()
+
+    # codec が無かったら StopIteration 例外が上がる
+    sendonly_codec_stats = next(
+        s for s in sendonly_stats if s.get("type") == "codec" and s.get("mimeType") == "audio/opus"
+    )
+    assert sendonly_codec_stats["mimeType"] == "audio/opus"
+
+    sendonly_codec_stats = next(
+        s for s in sendonly_stats if s.get("type") == "codec" and s.get("mimeType") == "video/VP9"
+    )
+    assert sendonly_codec_stats["mimeType"] == "video/VP9"
+
+    # outbound-rtp が無かったら StopIteration 例外が上がる
+    outbound_rtp_stats = next(
+        s for s in sendonly_stats if s.get("type") == "outbound-rtp" and s.get("kind") == "audio"
+    )
+    # audio には encoderImplementation が無い
+    assert outbound_rtp_stats["bytesSent"] > 0
+    assert outbound_rtp_stats["packetsSent"] > 0
+
+    # outbound-rtp が無かったら StopIteration 例外が上がる
+    outbound_rtp_stats = next(
+        s for s in sendonly_stats if s.get("type") == "outbound-rtp" and s.get("kind") == "video"
+    )
+    # video には encoderImplementation が無い
+    assert outbound_rtp_stats["bytesSent"] > 0
+    assert outbound_rtp_stats["packetsSent"] > 0
+
+    # codec が無かったら StopIteration 例外が上がる
+    recvonly_codec_stats = next(
+        s for s in recvonly_stats if s.get("type") == "codec" and s.get("mimeType") == "audio/opus"
+    )
+    assert recvonly_codec_stats["mimeType"] == "audio/opus"
+
+    recvonly_codec_stats = next(
+        s for s in recvonly_stats if s.get("type") == "codec" and s.get("mimeType") == "video/VP9"
+    )
+    assert recvonly_codec_stats["mimeType"] == "video/VP9"
+
+    # inbound-rtp が無かったら StopIteration 例外が上がる
+    inbound_rtp_stats = next(
+        s for s in recvonly_stats if s.get("type") == "inbound-rtp" and s.get("kind") == "audio"
+    )
+    # audio には encoderImplementation が無い
+    assert inbound_rtp_stats["bytesReceived"] > 0
+    assert inbound_rtp_stats["packetsReceived"] > 0
+
+    # inbound-rtp が無かったら StopIteration 例外が上がる
+    inbound_rtp_stats = next(
+        s for s in recvonly_stats if s.get("type") == "inbound-rtp" and s.get("kind") == "video"
+    )
+    # video には encoderImplementation が無い
+    assert inbound_rtp_stats["bytesReceived"] > 0
+    assert inbound_rtp_stats["packetsReceived"] > 0
+
+    # on_transform が呼ばれていることを確認
+    assert sendonly.is_called_on_audio_transform is True
+    assert sendonly.is_called_on_video_transform is True
+
+    assert recvonly.is_called_on_audio_transform is True
+    assert recvonly.is_called_on_video_transform is True