pipecat-ojin 0.1.1__tar.gz

pipecat_ojin-0.1.1/LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright [yyyy] [name of copyright owner]
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

pipecat_ojin-0.1.1/PKG-INFO

@@ -0,0 +1,107 @@
+Metadata-Version: 2.4
+Name: pipecat-ojin
+Version: 0.1.1
+Summary: Ojin avatar (Speech-To-Video) service for Pipecat
+Author: Ojin
+License: Apache-2.0
+Project-URL: Homepage, https://ojin.ai
+Project-URL: Repository, https://github.com/ojinai/pipecat-ojin
+Project-URL: Documentation, https://docs.ojin.ai
+Project-URL: Issues, https://github.com/ojinai/pipecat-ojin/issues
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Multimedia :: Sound/Audio
+Classifier: Topic :: Multimedia :: Video
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: pipecat-ai>=1.3.0
+Requires-Dist: ojin-client[stv]>=0.7.1
+Requires-Dist: pydantic<3,>=2
+Provides-Extra: dev
+Requires-Dist: pytest>=8.3.5; extra == "dev"
+Requires-Dist: pytest-asyncio>=0.26.0; extra == "dev"
+Requires-Dist: ruff>=0.11.4; extra == "dev"
+Requires-Dist: build>=0.10.0; extra == "dev"
+Dynamic: license-file
+
+# pipecat-ojin
+
+Ojin's [Pipecat](https://github.com/pipecat-ai/pipecat) integration: drop a
+**lip-synced talking-avatar face** (`OjinVideoService`) into your existing pipeline in minutes.
+
+`OjinVideoService` is the one stage that turns a voice agent into a video-call
+avatar — it lip-syncs to whatever your TTS produces and streams the avatar video
+back. It sits in the same slot as other video services in pipecat:
+
+```
+transport.input() -> STT -> LLM -> TTS -> [OjinVideoService] -> transport.output()
+```
+
+This package is a thin adapter over the framework-agnostic
+[`ojin-client`](https://github.com/ojinai/python-sdk) SDK — all avatar behaviour
+(A/V sync, audio-as-clock playback, barge-in re-sync) lives in the SDK. It is
+**not** a fork of Pipecat — it depends on `pipecat-ai` as a library.
+
+## Install
+
+```bash
+pip install pipecat-ojin
+```
+
+This pulls in `pipecat-ai` and `ojin-client[stv]`. You provide the STT / LLM /
+TTS services for your pipeline (e.g. `pip install "pipecat-ai[deepgram,groq,elevenlabs]"`).
+
+## Quickstart — the avatar face
+
+```python
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat_ojin import OjinVideoService, OjinVideoSettings
+
+avatar = OjinVideoService(
+    OjinVideoSettings(
+        api_key="OJIN_API_KEY",
+        config_id="OJIN_CONFIG_ID",   # the Face model to drive
+    )
+)
+
+pipeline = Pipeline(
+    [transport.input(), stt, llm, tts, avatar, transport.output()]
+)
+```
+
+The avatar's frame size comes from your Face model (`config_id`) — set your
+transport's `video_out_width` / `video_out_height` to match it (the example uses
+512×512).
+
+Get your `OJIN_API_KEY` and a Face model `OJIN_CONFIG_ID` from
+[ojin.ai](https://ojin.ai) (docs: [docs.ojin.ai](https://docs.ojin.ai)).
+
+### Session tracing (optional)
+
+Pass an `ojin.stv.OjinSessionTrace` to record a per-call Perfetto trace; the
+service dumps it on close:
+
+```python
+from ojin.stv import OjinSessionTrace
+
+trace = OjinSessionTrace(session_id="my-call", config_id="OJIN_CONFIG_ID")
+avatar = OjinVideoService(OjinVideoSettings(...), session_trace=trace)
+```
+
+## Example
+
+A complete, runnable voice + avatar agent (browser WebRTC or Daily) lives in
+[`examples/ojin-bot/`](examples/ojin-bot/).
+
+## Compatibility
+
+| Requirement | Version |
+|---|---|
+| Python | ≥ 3.11 |
+| `pipecat-ai` | ≥ 1.3.0 |
+| `ojin-client[stv]` | ≥ 0.7.1 |
+
+## License
+
+Apache-2.0. See [LICENSE](LICENSE).

pipecat_ojin-0.1.1/README.md

@@ -0,0 +1,80 @@
+# pipecat-ojin
+
+Ojin's [Pipecat](https://github.com/pipecat-ai/pipecat) integration: drop a
+**lip-synced talking-avatar face** (`OjinVideoService`) into your existing pipeline in minutes.
+
+`OjinVideoService` is the one stage that turns a voice agent into a video-call
+avatar — it lip-syncs to whatever your TTS produces and streams the avatar video
+back. It sits in the same slot as other video services in pipecat:
+
+```
+transport.input() -> STT -> LLM -> TTS -> [OjinVideoService] -> transport.output()
+```
+
+This package is a thin adapter over the framework-agnostic
+[`ojin-client`](https://github.com/ojinai/python-sdk) SDK — all avatar behaviour
+(A/V sync, audio-as-clock playback, barge-in re-sync) lives in the SDK. It is
+**not** a fork of Pipecat — it depends on `pipecat-ai` as a library.
+
+## Install
+
+```bash
+pip install pipecat-ojin
+```
+
+This pulls in `pipecat-ai` and `ojin-client[stv]`. You provide the STT / LLM /
+TTS services for your pipeline (e.g. `pip install "pipecat-ai[deepgram,groq,elevenlabs]"`).
+
+## Quickstart — the avatar face
+
+```python
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat_ojin import OjinVideoService, OjinVideoSettings
+
+avatar = OjinVideoService(
+    OjinVideoSettings(
+        api_key="OJIN_API_KEY",
+        config_id="OJIN_CONFIG_ID",   # the Face model to drive
+    )
+)
+
+pipeline = Pipeline(
+    [transport.input(), stt, llm, tts, avatar, transport.output()]
+)
+```
+
+The avatar's frame size comes from your Face model (`config_id`) — set your
+transport's `video_out_width` / `video_out_height` to match it (the example uses
+512×512).
+
+Get your `OJIN_API_KEY` and a Face model `OJIN_CONFIG_ID` from
+[ojin.ai](https://ojin.ai) (docs: [docs.ojin.ai](https://docs.ojin.ai)).
+
+### Session tracing (optional)
+
+Pass an `ojin.stv.OjinSessionTrace` to record a per-call Perfetto trace; the
+service dumps it on close:
+
+```python
+from ojin.stv import OjinSessionTrace
+
+trace = OjinSessionTrace(session_id="my-call", config_id="OJIN_CONFIG_ID")
+avatar = OjinVideoService(OjinVideoSettings(...), session_trace=trace)
+```
+
+## Example
+
+A complete, runnable voice + avatar agent (browser WebRTC or Daily) lives in
+[`examples/ojin-bot/`](examples/ojin-bot/).
+
+## Compatibility
+
+| Requirement | Version |
+|---|---|
+| Python | ≥ 3.11 |
+| `pipecat-ai` | ≥ 1.3.0 |
+| `ojin-client[stv]` | ≥ 0.7.1 |
+
+## License
+
+Apache-2.0. See [LICENSE](LICENSE).

pipecat_ojin-0.1.1/pyproject.toml

@@ -0,0 +1,58 @@
+[project]
+name = "pipecat-ojin"
+version = "0.1.1"
+description = "Ojin avatar (Speech-To-Video) service for Pipecat"
+readme = "README.md"
+# pipecat-ai (>=1.3.0) requires Python >=3.11, so the package floor matches it.
+requires-python = ">=3.11"
+authors = [{ name = "Ojin" }]
+license = { text = "Apache-2.0" }
+classifiers = [
+    "License :: OSI Approved :: Apache Software License",
+    "Programming Language :: Python :: 3",
+    "Topic :: Multimedia :: Sound/Audio",
+    "Topic :: Multimedia :: Video",
+]
+dependencies = [
+    "pipecat-ai>=1.3.0",
+    "ojin-client[stv]>=0.7.1",
+    "pydantic>=2,<3",
+]
+
+[build-system]
+requires = ["setuptools>=42", "wheel"]
+build-backend = "setuptools.build_meta"
+
+[project.urls]
+Homepage = "https://ojin.ai"
+Repository = "https://github.com/ojinai/pipecat-ojin"
+Documentation = "https://docs.ojin.ai"
+Issues = "https://github.com/ojinai/pipecat-ojin/issues"
+
+[project.optional-dependencies]
+dev = [
+    "pytest>=8.3.5",
+    "pytest-asyncio>=0.26.0",
+    "ruff>=0.11.4",
+    "build>=0.10.0",
+]
+
+[tool.setuptools]
+package-dir = { "" = "src" }
+packages = ["pipecat_ojin"]
+
+[tool.ruff]
+target-version = "py310"
+lint.select = ["E", "F", "I", "B", "SIM", "RUF"]
+lint.ignore = ["E501"]
+
+[tool.ruff.lint.isort]
+known-first-party = ["pipecat_ojin"]
+
+[tool.ruff.lint.per-file-ignores]
+"__init__.py" = ["F401"]
+
+[tool.pytest.ini_options]
+minversion = "8.0"
+testpaths = ["tests"]
+asyncio_mode = "auto"

pipecat_ojin-0.1.1/setup.cfg

@@ -0,0 +1,4 @@
+[egg_info]
+tag_build = 
+tag_date = 0
+

pipecat_ojin-0.1.1/src/pipecat_ojin/__init__.py

@@ -0,0 +1,20 @@
+"""pipecat-ojin: Ojin avatar (Speech-To-Video) service for Pipecat."""
+
+from pipecat_ojin.video import (
+    OjinBotStartedSpeakingFrame,
+    OjinBotStoppedSpeakingFrame,
+    OjinVideoInitializedFrame,
+    OjinVideoService,
+    OjinVideoSettings,
+)
+
+__version__ = "0.1.0"
+
+__all__ = [
+    "OjinBotStartedSpeakingFrame",
+    "OjinBotStoppedSpeakingFrame",
+    "OjinVideoInitializedFrame",
+    "OjinVideoService",
+    "OjinVideoSettings",
+    "__version__",
+]

pipecat_ojin-0.1.1/src/pipecat_ojin/video.py

@@ -0,0 +1,253 @@
+"""OjinVideoService — a Pipecat adapter for an Ojin lip-synced talking avatar.
+
+`OjinVideoService` is a Pipecat ``FrameProcessor`` that turns the TTS audio
+stream into a lip-synced avatar. Place it after your TTS service and before
+``transport.output()`` — the slot Pipecat's built-in avatar services use::
+
+    transport.input() -> STT -> LLM -> TTS -> [OjinVideoService] -> transport.output()
+
+All avatar behaviour (connect/retry, audio-as-clock playback, A/V sync, re-sync
+after barge-in, off-loop JPEG decode, session tracing) lives in
+``ojin.stv.OjinSTVClient``; this class is only the mapping between Pipecat frames
+and the client's API. It needs ``ojin-client[stv]`` and ``pipecat-ai``.
+"""
+
+from __future__ import annotations
+
+import logging
+import os
+from dataclasses import dataclass
+from datetime import datetime, timezone
+from typing import Optional
+
+from ojin.stv import (
+    OjinSessionTrace,
+    OjinSTVClient,
+    STVAudioFrame,
+    STVEvent,
+    STVVideoFrame,
+)
+from pipecat.frames.frames import (
+    CancelFrame,
+    EndFrame,
+    Frame,
+    InterruptionFrame,
+    OutputAudioRawFrame,
+    OutputImageRawFrame,
+    StartFrame,
+    TTSAudioRawFrame,
+    TTSStartedFrame,
+    UserStartedSpeakingFrame,
+)
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+logger = logging.getLogger(__name__)
+
+# Per-session Perfetto trace output dir (same layout as the python-sdk example).
+_TRACE_ROOT = os.getenv("OJIN_STV_TRACE_DIR", "/root/debug/sessions/stv-pipecat-ojin")
+
+# The ~0.5 s all-zero trailing-silence sentinel some TTS engines emit. These mirror
+# OjinSTVClient.send_tts_audio's own discard rule (ojin.stv) so the adapter can drop
+# it *before* arming TTFB — TTFB must time the first real audio, not the silence.
+_HALF_SECOND = 0.5
+_HALF_SECOND_TOL = 0.01
+
+
+@dataclass
+class OjinVideoInitializedFrame(Frame):
+    """Frame indicating the avatar session is ready (server handshake complete)."""
+
+    session_data: Optional[dict] = None
+
+
+class OjinBotStartedSpeakingFrame(Frame):
+    """Emitted when the avatar starts speaking (a buffer is promoted to current)."""
+
+    pass
+
+
+class OjinBotStoppedSpeakingFrame(Frame):
+    """Emitted when the avatar stops speaking (current buffer drains, none queued)."""
+
+    pass
+
+
+@dataclass
+class OjinVideoSettings:
+    """Connection identity and frame size for :class:`OjinVideoService`."""
+
+    api_key: str = ""
+    config_id: str = ""
+    ws_url: str = "wss://models.ojin.ai/realtime"
+    # When True (default), TTS audio sent during the avatar's cold-start handshake
+    # is buffered and replayed once the session is ready, instead of dropped — so
+    # an opening line isn't lost. Set False to drop pre-init audio (old behavior).
+    buffer_preinit_tts_audio: bool = True
+
+
+class _PushFrameOutput:
+    """STVOutput sink: forwards the client's synced A/V downstream, behind the gate."""
+
+    def __init__(self, service: "OjinVideoService") -> None:
+        self._svc = service
+
+    async def write_audio(self, frame: STVAudioFrame) -> None:
+        if self._svc._can_start_playback:
+            await self._svc.push_frame(
+                OutputAudioRawFrame(frame.pcm, frame.sample_rate, frame.num_channels)
+            )
+
+    async def write_video(self, frame: STVVideoFrame) -> None:
+        if self._svc._can_start_playback and frame.rgb is not None:
+            await self._svc.push_frame(
+                OutputImageRawFrame(
+                    image=frame.rgb,
+                    size=(frame.width, frame.height),
+                    format=frame.format,
+                )
+            )
+
+    def on_event(self, event: STVEvent, **kwargs: object) -> None:
+        pass  # lifecycle events are wired via the client's emitter (see _wire_events)
+
+
+class OjinVideoService(FrameProcessor):
+    """Pipecat FrameProcessor that turns the TTS audio stream into a lip-synced avatar."""
+
+    def __init__(
+        self,
+        settings: OjinVideoSettings,
+        *,
+        session_trace: Optional[OjinSessionTrace] = None,
+        stv_client: Optional[OjinSTVClient] = None,
+    ) -> None:
+        """Build the adapter and wire the client's lifecycle events.
+
+        Args:
+            settings: connection identity + avatar frame size.
+            session_trace: an ``ojin.stv.OjinSessionTrace`` to record this call;
+                injected as the client's tracer and dumped to disk on close.
+                ``None`` -> the client uses a no-op tracer.
+            stv_client: a pre-built client (dependency injection for tests or
+                alternative transports). When given, ``settings`` connection
+                fields are not used to build a client.
+        """
+        super().__init__(name="ojin-video")
+        self._settings = settings
+        self._can_start_playback = True  # gate open by default; close to defer A/V
+        self._waiting_for_first_tts = False
+        self._trace = session_trace
+        self._output = _PushFrameOutput(self)
+        self._stv = stv_client or OjinSTVClient(
+            api_key=settings.api_key,
+            config_id=settings.config_id,
+            ws_url=settings.ws_url,
+            output=self._output,
+            tracer=session_trace,
+            buffer_preinit_tts_audio=settings.buffer_preinit_tts_audio,
+        )
+        self._wire_events()
+
+    def set_can_start_playback(self, value: bool) -> None:
+        """Open (``True``) or close (``False``) the playback gate.
+
+        Defaults open. Close it before participant join to keep the connect->join
+        idle backlog out of the transport, then re-open at join.
+        """
+        self._can_start_playback = value
+
+    def can_generate_metrics(self) -> bool:
+        """Enable Pipecat TTFB / processing metrics for this service."""
+        return True
+
+    async def connect_with_retry(self) -> bool:
+        """Connect the underlying client with retry; ``True`` on success."""
+        return await self._stv.connect_with_retry()
+
+    def _wire_events(self) -> None:
+        """Map ``OjinSTVClient`` events onto Pipecat frames + TTFB metrics."""
+
+        @self._stv.on(STVEvent.SESSION_READY)
+        async def _on_ready(session_data=None, **_):
+            frame = OjinVideoInitializedFrame(session_data=session_data)
+            await self.push_frame(frame, FrameDirection.DOWNSTREAM)
+            await self.push_frame(frame, FrameDirection.UPSTREAM)
+
+        @self._stv.on(STVEvent.BOT_STARTED_SPEAKING)
+        async def _on_started(**_):
+            await self.push_frame(OjinBotStartedSpeakingFrame())
+            await self.stop_ttfb_metrics()
+
+        @self._stv.on(STVEvent.BOT_STOPPED_SPEAKING)
+        async def _on_stopped(**_):
+            await self.push_frame(OjinBotStoppedSpeakingFrame())
+
+        @self._stv.on(STVEvent.ERROR)
+        async def _on_error(message="", fatal=False, **_):
+            await self.push_error(message, fatal=fatal)
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection) -> None:
+        """Map each inbound Pipecat frame to the matching OjinSTVClient call."""
+        await super().process_frame(frame, direction)
+
+        if isinstance(frame, StartFrame):
+            await self.push_frame(frame, direction)
+            await self._stv.start()
+        elif isinstance(frame, TTSStartedFrame):
+            self._waiting_for_first_tts = True
+            await self._stv.start_turn()
+            await self.push_frame(frame, direction)
+        elif isinstance(frame, TTSAudioRawFrame):
+            if self._is_trailing_silence(frame):
+                # Drop the trailing-silence sentinel here, before arming TTFB: TTFB
+                # must time the first *real* audio. It never reaches the client (which
+                # would discard it too) and leaves TTFB un-armed for the next frame.
+                return
+            if self._waiting_for_first_tts:
+                self._waiting_for_first_tts = False
+                await self.start_ttfb_metrics()
+            await self._stv.send_tts_audio(
+                frame.audio, frame.sample_rate, frame.num_channels
+            )
+        elif isinstance(frame, (InterruptionFrame, UserStartedSpeakingFrame)):
+            # Barge-in: an explicit interruption — or the user starting to speak —
+            # cuts the avatar's current turn. Forward the frame so the rest of the
+            # pipeline still sees it.
+            await self._stv.interrupt()
+            await self.push_frame(frame, direction)
+        elif isinstance(frame, (EndFrame, CancelFrame)):
+            await self._stv.close()
+            self._write_trace()
+            await self.push_frame(frame, direction)
+        else:
+            await self.push_frame(frame, direction)
+
+    @staticmethod
+    def _is_trailing_silence(frame: TTSAudioRawFrame) -> bool:
+        """Whether ``frame`` is the ~0.5 s all-zero trailing-silence sentinel.
+
+        Mirrors ``OjinSTVClient.send_tts_audio``'s discard rule so the adapter can
+        drop it before arming TTFB metrics (which must time the first real audio).
+        """
+        pcm = frame.audio
+        if not pcm or any(pcm):  # empty, or any non-zero byte -> real audio
+            return False
+        duration = len(pcm) / (frame.sample_rate * frame.num_channels * 2)
+        return abs(duration - _HALF_SECOND) < _HALF_SECOND_TOL
+
+    def _write_trace(self) -> None:
+        """Dump the session's Perfetto trace to disk on close (best-effort, once)."""
+        trace, self._trace = self._trace, None  # write at most once
+        if trace is None:
+            return
+        now = datetime.now(timezone.utc)
+        path = os.path.join(
+            _TRACE_ROOT,
+            now.strftime("%Y-%m-%d"),
+            f"{now.strftime('%H-%M-%S')}_{trace.session_id}",
+            "session.json",
+        )
+        try:
+            logger.info("Ojin STV session trace written to %s", trace.dump(path))
+        except Exception as exc:  # never let trace I/O break teardown
+            logger.warning("Failed to write Ojin STV session trace: %s", exc)

pipecat_ojin-0.1.1/src/pipecat_ojin.egg-info/PKG-INFO

@@ -0,0 +1,107 @@
+Metadata-Version: 2.4
+Name: pipecat-ojin
+Version: 0.1.1
+Summary: Ojin avatar (Speech-To-Video) service for Pipecat
+Author: Ojin
+License: Apache-2.0
+Project-URL: Homepage, https://ojin.ai
+Project-URL: Repository, https://github.com/ojinai/pipecat-ojin
+Project-URL: Documentation, https://docs.ojin.ai
+Project-URL: Issues, https://github.com/ojinai/pipecat-ojin/issues
+Classifier: License :: OSI Approved :: Apache Software License
+Classifier: Programming Language :: Python :: 3
+Classifier: Topic :: Multimedia :: Sound/Audio
+Classifier: Topic :: Multimedia :: Video
+Requires-Python: >=3.11
+Description-Content-Type: text/markdown
+License-File: LICENSE
+Requires-Dist: pipecat-ai>=1.3.0
+Requires-Dist: ojin-client[stv]>=0.7.1
+Requires-Dist: pydantic<3,>=2
+Provides-Extra: dev
+Requires-Dist: pytest>=8.3.5; extra == "dev"
+Requires-Dist: pytest-asyncio>=0.26.0; extra == "dev"
+Requires-Dist: ruff>=0.11.4; extra == "dev"
+Requires-Dist: build>=0.10.0; extra == "dev"
+Dynamic: license-file
+
+# pipecat-ojin
+
+Ojin's [Pipecat](https://github.com/pipecat-ai/pipecat) integration: drop a
+**lip-synced talking-avatar face** (`OjinVideoService`) into your existing pipeline in minutes.
+
+`OjinVideoService` is the one stage that turns a voice agent into a video-call
+avatar — it lip-syncs to whatever your TTS produces and streams the avatar video
+back. It sits in the same slot as other video services in pipecat:
+
+```
+transport.input() -> STT -> LLM -> TTS -> [OjinVideoService] -> transport.output()
+```
+
+This package is a thin adapter over the framework-agnostic
+[`ojin-client`](https://github.com/ojinai/python-sdk) SDK — all avatar behaviour
+(A/V sync, audio-as-clock playback, barge-in re-sync) lives in the SDK. It is
+**not** a fork of Pipecat — it depends on `pipecat-ai` as a library.
+
+## Install
+
+```bash
+pip install pipecat-ojin
+```
+
+This pulls in `pipecat-ai` and `ojin-client[stv]`. You provide the STT / LLM /
+TTS services for your pipeline (e.g. `pip install "pipecat-ai[deepgram,groq,elevenlabs]"`).
+
+## Quickstart — the avatar face
+
+```python
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat_ojin import OjinVideoService, OjinVideoSettings
+
+avatar = OjinVideoService(
+    OjinVideoSettings(
+        api_key="OJIN_API_KEY",
+        config_id="OJIN_CONFIG_ID",   # the Face model to drive
+    )
+)
+
+pipeline = Pipeline(
+    [transport.input(), stt, llm, tts, avatar, transport.output()]
+)
+```
+
+The avatar's frame size comes from your Face model (`config_id`) — set your
+transport's `video_out_width` / `video_out_height` to match it (the example uses
+512×512).
+
+Get your `OJIN_API_KEY` and a Face model `OJIN_CONFIG_ID` from
+[ojin.ai](https://ojin.ai) (docs: [docs.ojin.ai](https://docs.ojin.ai)).
+
+### Session tracing (optional)
+
+Pass an `ojin.stv.OjinSessionTrace` to record a per-call Perfetto trace; the
+service dumps it on close:
+
+```python
+from ojin.stv import OjinSessionTrace
+
+trace = OjinSessionTrace(session_id="my-call", config_id="OJIN_CONFIG_ID")
+avatar = OjinVideoService(OjinVideoSettings(...), session_trace=trace)
+```
+
+## Example
+
+A complete, runnable voice + avatar agent (browser WebRTC or Daily) lives in
+[`examples/ojin-bot/`](examples/ojin-bot/).
+
+## Compatibility
+
+| Requirement | Version |
+|---|---|
+| Python | ≥ 3.11 |
+| `pipecat-ai` | ≥ 1.3.0 |
+| `ojin-client[stv]` | ≥ 0.7.1 |
+
+## License
+
+Apache-2.0. See [LICENSE](LICENSE).

pipecat_ojin-0.1.1/src/pipecat_ojin.egg-info/SOURCES.txt

@@ -0,0 +1,12 @@
+LICENSE
+README.md
+pyproject.toml
+src/pipecat_ojin/__init__.py
+src/pipecat_ojin/video.py
+src/pipecat_ojin.egg-info/PKG-INFO
+src/pipecat_ojin.egg-info/SOURCES.txt
+src/pipecat_ojin.egg-info/dependency_links.txt
+src/pipecat_ojin.egg-info/requires.txt
+src/pipecat_ojin.egg-info/top_level.txt
+tests/test_smoke.py
+tests/test_video.py

pipecat_ojin-0.1.1/src/pipecat_ojin.egg-info/dependency_links.txt

@@ -0,0 +1 @@
+

pipecat_ojin-0.1.1/src/pipecat_ojin.egg-info/requires.txt

@@ -0,0 +1,9 @@
+pipecat-ai>=1.3.0
+ojin-client[stv]>=0.7.1
+pydantic<3,>=2
+
+[dev]
+pytest>=8.3.5
+pytest-asyncio>=0.26.0
+ruff>=0.11.4
+build>=0.10.0

pipecat_ojin-0.1.1/src/pipecat_ojin.egg-info/top_level.txt

@@ -0,0 +1 @@
+pipecat_ojin

pipecat_ojin-0.1.1/tests/test_smoke.py

@@ -0,0 +1,7 @@
+"""Smoke test: the package installs and imports with the expected version."""
+
+
+def test_package_imports_with_version():
+    import pipecat_ojin
+
+    assert pipecat_ojin.__version__ == "0.1.0"

pipecat_ojin-0.1.1/tests/test_video.py

@@ -0,0 +1,361 @@
+"""Tests for the OjinVideoService adapter over ojin.stv.OjinSTVClient.
+
+The adapter only (a) translates Pipecat frames into OjinSTVClient calls,
+(b) implements the STVOutput sink (pushing Output*RawFrame downstream) behind
+the playback-start gate, and (c) maps client events to Pipecat frames + TTFB
+metrics. All avatar behaviour lives in ojin.stv and is tested there.
+"""
+
+import unittest
+from unittest.mock import AsyncMock
+
+from ojin.stv import STVAudioFrame, STVEvent, STVVideoFrame
+from ojin.stv.events import EventEmitter
+from pipecat.frames.frames import (
+    CancelFrame,
+    EndFrame,
+    OutputAudioRawFrame,
+    OutputImageRawFrame,
+    TTSAudioRawFrame,
+    TTSStartedFrame,
+    UserStartedSpeakingFrame,
+)
+from pipecat.processors.frame_processor import FrameDirection
+
+from pipecat_ojin.video import (
+    OjinBotStartedSpeakingFrame,
+    OjinBotStoppedSpeakingFrame,
+    OjinVideoInitializedFrame,
+    OjinVideoService,
+    OjinVideoSettings,
+)
+
+try:
+    from pipecat.tests.utils import run_test
+
+    _HAS_RUN_TEST = True
+except Exception:  # pragma: no cover - depends on pipecat packaging test utils
+    _HAS_RUN_TEST = False
+
+
+class FakeSTVClient:
+    """Stand-in for OjinSTVClient.
+
+    Records adapter->client calls and uses the real EventEmitter so the adapter's
+    event wiring runs against production dispatch logic.
+    """
+
+    def __init__(self) -> None:
+        self._events = EventEmitter()
+        self.calls: list = []
+        self.connect_return = True
+
+    def on(self, event):
+        return self._events.on(event)
+
+    def add_listener(self, event, cb) -> None:
+        self._events.add_listener(event, cb)
+
+    async def emit(self, event, **kwargs) -> None:
+        await self._events.emit(event, **kwargs)
+
+    async def start(self) -> None:
+        self.calls.append("start")
+
+    async def start_turn(self) -> None:
+        self.calls.append("start_turn")
+
+    async def send_tts_audio(self, pcm, sample_rate, num_channels) -> None:
+        self.calls.append(("send_tts_audio", pcm, sample_rate, num_channels))
+
+    async def interrupt(self) -> None:
+        self.calls.append("interrupt")
+
+    async def close(self) -> None:
+        self.calls.append("close")
+
+    async def connect_with_retry(self) -> bool:
+        self.calls.append("connect_with_retry")
+        return self.connect_return
+
+
+def _audio() -> TTSAudioRawFrame:
+    return TTSAudioRawFrame(audio=b"\x01\x00" * 160, sample_rate=24000, num_channels=1)
+
+
+def _silence() -> TTSAudioRawFrame:
+    # 0.5 s of all-zero PCM @ 24 kHz mono int16 == the client's trailing-silence
+    # sentinel (24000 samples * 2 bytes = 24000 bytes => 0.5 s).
+    return TTSAudioRawFrame(audio=b"\x00" * 24000, sample_rate=24000, num_channels=1)
+
+
+def _adapter(fake: FakeSTVClient, **settings_kw) -> OjinVideoService:
+    return OjinVideoService(OjinVideoSettings(**settings_kw), stv_client=fake)
+
+
+class TestPlaybackGate(unittest.IsolatedAsyncioTestCase):
+    """The gate lives in the adapter; default open, close to drop A/V."""
+
+    async def test_open_by_default_forwards_audio_and_video(self) -> None:
+        svc = _adapter(FakeSTVClient())
+        svc.push_frame = AsyncMock()
+        await svc._output.write_audio(
+            STVAudioFrame(pcm=b"\x02\x00", sample_rate=24000, num_channels=1, pts=0)
+        )
+        await svc._output.write_video(
+            STVVideoFrame(
+                rgb=b"rgbrgb",
+                source_bytes=b"jpg",
+                width=1,
+                height=2,
+                frame_type=1,
+                pts=0,
+            )
+        )
+        pushed = [c.args[0] for c in svc.push_frame.call_args_list]
+        audio = [f for f in pushed if isinstance(f, OutputAudioRawFrame)]
+        image = [f for f in pushed if isinstance(f, OutputImageRawFrame)]
+        self.assertEqual(len(audio), 1)
+        self.assertEqual(audio[0].audio, b"\x02\x00")
+        self.assertEqual(audio[0].sample_rate, 24000)
+        self.assertEqual(audio[0].num_channels, 1)
+        self.assertEqual(len(image), 1)
+        self.assertEqual(image[0].image, b"rgbrgb")
+        self.assertEqual(image[0].size, (1, 2))
+        self.assertEqual(image[0].format, "RGB")
+
+    async def test_closed_gate_drops_audio_and_video(self) -> None:
+        svc = _adapter(FakeSTVClient())
+        svc.push_frame = AsyncMock()
+        svc.set_can_start_playback(False)
+        await svc._output.write_audio(
+            STVAudioFrame(pcm=b"\x01\x00", sample_rate=24000, num_channels=1, pts=0)
+        )
+        await svc._output.write_video(
+            STVVideoFrame(
+                rgb=b"rgb", source_bytes=b"jpg", width=4, height=4, frame_type=1, pts=0
+            )
+        )
+        svc.push_frame.assert_not_called()
+
+    async def test_open_but_no_rgb_drops_video(self) -> None:
+        svc = _adapter(FakeSTVClient())
+        svc.push_frame = AsyncMock()
+        await svc._output.write_video(
+            STVVideoFrame(
+                rgb=None, source_bytes=b"jpg", width=1, height=1, frame_type=0, pts=0
+            )
+        )
+        svc.push_frame.assert_not_called()
+
+
+class TestEventToFrameMapping(unittest.IsolatedAsyncioTestCase):
+    """Client lifecycle events map to the Pipecat frames + TTFB the bot expects."""
+
+    def _wired(self):
+        fake = FakeSTVClient()
+        svc = _adapter(fake)
+        svc.push_frame = AsyncMock()
+        svc.push_error = AsyncMock()
+        svc.start_ttfb_metrics = AsyncMock()
+        svc.stop_ttfb_metrics = AsyncMock()
+        return fake, svc
+
+    async def test_session_ready_pushes_initialized_frame_both_directions(self) -> None:
+        fake, svc = self._wired()
+        await fake.emit(STVEvent.SESSION_READY, session_data={"foo": 1})
+        init = [
+            c
+            for c in svc.push_frame.call_args_list
+            if isinstance(c.args[0], OjinVideoInitializedFrame)
+        ]
+        self.assertEqual(len(init), 2)
+        self.assertEqual(init[0].args[0].session_data, {"foo": 1})
+        dirs = {c.args[1] for c in init}
+        self.assertEqual(dirs, {FrameDirection.DOWNSTREAM, FrameDirection.UPSTREAM})
+
+    async def test_started_speaking_emits_frame_and_stops_ttfb(self) -> None:
+        fake, svc = self._wired()
+        await fake.emit(STVEvent.BOT_STARTED_SPEAKING)
+        pushed = [type(c.args[0]) for c in svc.push_frame.call_args_list]
+        self.assertIn(OjinBotStartedSpeakingFrame, pushed)
+        svc.stop_ttfb_metrics.assert_awaited_once()
+
+    async def test_stopped_speaking_emits_frame(self) -> None:
+        fake, svc = self._wired()
+        await fake.emit(STVEvent.BOT_STOPPED_SPEAKING)
+        pushed = [type(c.args[0]) for c in svc.push_frame.call_args_list]
+        self.assertIn(OjinBotStoppedSpeakingFrame, pushed)
+
+    async def test_error_event_pushes_error_with_message_and_fatal(self) -> None:
+        fake, svc = self._wired()
+        await fake.emit(STVEvent.ERROR, message="boom", code="X", fatal=True)
+        svc.push_error.assert_awaited_once()
+        call = svc.push_error.call_args
+        self.assertEqual(call.args[0], "boom")
+        self.assertTrue(call.kwargs.get("fatal"))
+
+    async def test_error_event_without_code_kwarg(self) -> None:
+        fake, svc = self._wired()
+        await fake.emit(STVEvent.ERROR, message="connect failed", fatal=True)
+        svc.push_error.assert_awaited_once()
+        self.assertEqual(svc.push_error.call_args.args[0], "connect failed")
+
+
+class TestFrameRouting(unittest.IsolatedAsyncioTestCase):
+    """Inbound Pipecat frames route to the right client calls (mid-stream frames)."""
+
+    def _svc(self, **kw):
+        fake = FakeSTVClient()
+        svc = _adapter(fake, **kw)
+        svc.push_frame = AsyncMock()
+        svc.start_ttfb_metrics = AsyncMock()
+        return fake, svc
+
+    async def test_tts_started_opens_turn(self) -> None:
+        fake, svc = self._svc()
+        await svc.process_frame(TTSStartedFrame(), FrameDirection.DOWNSTREAM)
+        self.assertIn("start_turn", fake.calls)
+
+    async def test_tts_audio_sends_to_client_and_arms_ttfb_once(self) -> None:
+        fake, svc = self._svc()
+        await svc.process_frame(TTSStartedFrame(), FrameDirection.DOWNSTREAM)
+        first = _audio()
+        await svc.process_frame(first, FrameDirection.DOWNSTREAM)
+        await svc.process_frame(_audio(), FrameDirection.DOWNSTREAM)
+        sends = [
+            c for c in fake.calls if isinstance(c, tuple) and c[0] == "send_tts_audio"
+        ]
+        self.assertEqual(len(sends), 2)
+        self.assertEqual(
+            sends[0],
+            ("send_tts_audio", first.audio, first.sample_rate, first.num_channels),
+        )
+        svc.start_ttfb_metrics.assert_awaited_once()
+
+    async def test_tts_audio_not_pushed_downstream(self) -> None:
+        # The adapter never passes the TTS audio frame through; the client's
+        # output sink emits the played audio instead.
+        _fake, svc = self._svc()
+        frame = _audio()
+        await svc.process_frame(frame, FrameDirection.DOWNSTREAM)
+        pushed = [c.args[0] for c in svc.push_frame.call_args_list]
+        self.assertNotIn(frame, pushed)
+
+    async def test_user_started_speaking_interrupts(self) -> None:
+        fake, svc = self._svc()
+        await svc.process_frame(UserStartedSpeakingFrame(), FrameDirection.DOWNSTREAM)
+        self.assertIn("interrupt", fake.calls)
+
+    async def test_end_frame_closes_client(self) -> None:
+        fake, svc = self._svc()
+        await svc.process_frame(EndFrame(), FrameDirection.DOWNSTREAM)
+        self.assertIn("close", fake.calls)
+
+    async def test_cancel_frame_closes_client(self) -> None:
+        fake, svc = self._svc()
+        await svc.process_frame(CancelFrame(), FrameDirection.DOWNSTREAM)
+        self.assertIn("close", fake.calls)
+
+    async def test_can_generate_metrics_is_true(self) -> None:
+        _, svc = self._svc()
+        self.assertTrue(svc.can_generate_metrics())
+
+
+class TestTtfbAndSilence(unittest.IsolatedAsyncioTestCase):
+    """TTFB arming honours the trailing-silence sentinel and re-arms per turn."""
+
+    def _svc(self, **kw):
+        fake = FakeSTVClient()
+        svc = _adapter(fake, **kw)
+        svc.push_frame = AsyncMock()
+        svc.start_ttfb_metrics = AsyncMock()
+        return fake, svc
+
+    async def test_trailing_silence_first_frame_is_dropped_not_armed(self) -> None:
+        fake, svc = self._svc()
+        await svc.process_frame(TTSStartedFrame(), FrameDirection.DOWNSTREAM)
+        await svc.process_frame(_silence(), FrameDirection.DOWNSTREAM)
+        svc.start_ttfb_metrics.assert_not_awaited()
+        self.assertNotIn(
+            "send_tts_audio", [c[0] for c in fake.calls if isinstance(c, tuple)]
+        )
+        # the real first frame that follows still arms TTFB once
+        await svc.process_frame(_audio(), FrameDirection.DOWNSTREAM)
+        svc.start_ttfb_metrics.assert_awaited_once()
+
+    async def test_ttfb_not_armed_without_a_tts_started_frame(self) -> None:
+        _, svc = self._svc()
+        await svc.process_frame(_audio(), FrameDirection.DOWNSTREAM)
+        svc.start_ttfb_metrics.assert_not_awaited()
+
+    async def test_ttfb_rearmed_on_each_turn(self) -> None:
+        _, svc = self._svc()
+        for _turn in range(2):
+            await svc.process_frame(TTSStartedFrame(), FrameDirection.DOWNSTREAM)
+            await svc.process_frame(_audio(), FrameDirection.DOWNSTREAM)
+        self.assertEqual(svc.start_ttfb_metrics.await_count, 2)
+
+
+class TestClientDelegation(unittest.IsolatedAsyncioTestCase):
+    """connect_with_retry delegates to the client."""
+
+    async def test_connect_with_retry_delegates_and_returns_true(self) -> None:
+        fake = FakeSTVClient()
+        fake.connect_return = True
+        svc = _adapter(fake)
+        self.assertTrue(await svc.connect_with_retry())
+        self.assertIn("connect_with_retry", fake.calls)
+
+    async def test_connect_with_retry_propagates_false(self) -> None:
+        fake = FakeSTVClient()
+        fake.connect_return = False
+        svc = _adapter(fake)
+        self.assertFalse(await svc.connect_with_retry())
+
+
+class TestFrameTransparency(unittest.IsolatedAsyncioTestCase):
+    """Handled frames are still forwarded; unknown frames pass through."""
+
+    def _svc(self, **kw):
+        fake = FakeSTVClient()
+        svc = _adapter(fake, **kw)
+        svc.push_frame = AsyncMock()
+        svc.start_ttfb_metrics = AsyncMock()
+        return fake, svc
+
+    def _pushed_args(self, svc):
+        return [c.args for c in svc.push_frame.call_args_list]
+
+    async def test_tts_started_forwarded_downstream(self) -> None:
+        _fake, svc = self._svc()
+        frame = TTSStartedFrame()
+        await svc.process_frame(frame, FrameDirection.DOWNSTREAM)
+        self.assertIn((frame, FrameDirection.DOWNSTREAM), self._pushed_args(svc))
+
+    async def test_unknown_frame_passes_through(self) -> None:
+        _fake, svc = self._svc()
+        frame = OutputAudioRawFrame(b"\x00\x00", 24000, 1)
+        await svc.process_frame(frame, FrameDirection.DOWNSTREAM)
+        self.assertIn((frame, FrameDirection.DOWNSTREAM), self._pushed_args(svc))
+
+
+@unittest.skipUnless(_HAS_RUN_TEST, "pipecat.tests.utils.run_test not available")
+class TestLifecycleThroughPipeline(unittest.IsolatedAsyncioTestCase):
+    """End-to-end through a real pipeline: StartFrame -> start, EndFrame -> close."""
+
+    async def test_start_starts_client_and_end_closes_it(self) -> None:
+        fake = FakeSTVClient()
+        svc = _adapter(fake)
+        await run_test(
+            svc,
+            frames_to_send=[],
+            expected_down_frames=None,
+            send_end_frame=True,
+        )
+        self.assertIn("start", fake.calls)
+        self.assertIn("close", fake.calls)
+
+
+if __name__ == "__main__":
+    unittest.main()