Add a currency test to test_utils_string unit test

GladiaSTTService: deprecate confidence arg

CHANGELOG.md

@@ -9,6 +9,11 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Added
 
+- Added support for global location in `GoogleVertexLLMService`. The service now
+  supports both regional locations (e.g., "us-east4") and the "global" location
+  for Vertex AI endpoints. When using "global" location, the service will use
+  `aiplatform.googleapis.com` as the API host instead of the regional format.
+
 - Added `on_pipeline_finished` event to `PipelineTask`. This event will get
   fired when the pipeline is done running. This can be the result of a
   `StopFrame`, `CancelFrame` or `EndFrame`.
@@ -19,14 +24,60 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
       ...
   ```
 
+### Changed
+
+- Updated `livekit` to 1.0.13.
+
+- `torch` and `torchaudio` are no longer required for running Smart Turn
+  locally. This avoids gigabytes of dependencies being installed.
+
+- Updated `websockets` dependency to support version 15.0. Removed deprecated
+  usage of `ConnectionClosed.code` and `ConnectionClosed.reason` attributes in
+  `AWSTranscribeSTTService` for compatibility.
+
+- Refactored `pyproject.toml` to reduce websockets dependency repetition using
+  self-referencing extras. All websockets-dependent services now reference a
+  shared `websockets-base` extra.
+
 ### Deprecated
 
+- `GladiaSTTService`'s `confidence` arg is deprecated. `confidence` is no
+  longer needed to determine which transcription or translation frames to
+  emit.
+
 - `PipelineTask` events `on_pipeline_stopped`, `on_pipeline_ended` and
   `on_pipeline_cancelled` are now deprecated. Use `on_pipeline_finished`
   instead.
 
 ### Fixed
 
+- Fixed `DailyTransport.sip_call_transfer()` to automatically use the session
+  ID from the `on_dialin_connected` event, when not explicitly provided. Now
+  supports cold transfers (from incoming dial-in calls) by automatically
+  tracking session IDs from connection events.
+
+- Fixed a memory leak in `SmallWebRTCTransport`. In `aiortc`, when you receive
+  a `MediaStreamTrack` (audio or video), frames are produced asynchronously. If
+  the code never consumes these frames, they are queued in memory, causing a
+  memory leak.
+
+- Fixed an issue in `AsyncAITTSService`, where `TTSTextFrames` were not being
+  pushed.
+
+- Fixed an issue that would cause `push_interruption_task_frame_and_wait()` to
+  not wait if a previous interruption had already happened.
+
+- Fixed a couple of bugs in `ServiceSwitcher`:
+
+  - Using multiple `ServiceSwitcher`s in a pipeline would result in an error.
+  - `ServiceSwitcherFrame`s (such as `ManuallySwitchServiceFrame`s) were having
+    an effect too early, essentially "jumping the queue" in terms of pipeline
+    frame ordering.
+
+- Fixed a self-cancellation deadlock in `UserIdleProcessor` when returning
+  `False` from an idle callback. The task now terminates naturally instead of
+  attempting to cancel itself.
+
 - Fixed an issue in `AudioBufferProcessor` where a recording is not created
   when a bot speaks and user input is blocked.
 

README.md

@@ -21,6 +21,8 @@
 
 🧭 Looking to build structured conversations? Check out [Pipecat Flows](https://github.com/pipecat-ai/pipecat-flows) for managing complex conversational states and transitions.
 
+🔍 Looking for help debugging your pipeline and processors? Check out [Whisker](https://github.com/pipecat-ai/whisker), a real-time Pipecat debugger.
+
 ## 🧠 Why Pipecat?
 
 - **Voice-first**: Integrates speech recognition, text-to-speech, and conversation handling

examples/foundational/01b-livekit-audio.py

@@ -11,7 +11,7 @@ import sys
 from dotenv import load_dotenv
 from loguru import logger
 
-from pipecat.frames.frames import TextFrame
+from pipecat.frames.frames import TTSSpeakFrame
 from pipecat.pipeline.pipeline import Pipeline
 from pipecat.pipeline.runner import PipelineRunner
 from pipecat.pipeline.task import PipelineTask
@@ -50,7 +50,7 @@ async def main():
     async def on_first_participant_joined(transport, participant_id):
         await asyncio.sleep(1)
         await task.queue_frame(
-            TextFrame(
+            TTSSpeakFrame(
                 "Hello there! How are you doing today? Would you like to talk about the weather?"
             )
         )

examples/foundational/38b-smart-turn-local.py

@@ -30,10 +30,6 @@ from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
 
 load_dotenv(override=True)
 
-# To use this locally, set the environment variable LOCAL_SMART_TURN_MODEL_PATH
-# to the Smart Turn v3 ONNX model file.
-smart_turn_model_path = os.getenv("LOCAL_SMART_TURN_MODEL_PATH")
-
 # We store functions so objects (e.g. SileroVADAnalyzer) don't get
 # instantiated. The function will be called when the desired transport gets
 # selected.
@@ -42,25 +38,19 @@ transport_params = {
         audio_in_enabled=True,
         audio_out_enabled=True,
         vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(
-            smart_turn_model_path=smart_turn_model_path, params=SmartTurnParams()
-        ),
+        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "twilio": lambda: FastAPIWebsocketParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
         vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(
-            smart_turn_model_path=smart_turn_model_path, params=SmartTurnParams()
-        ),
+        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
     "webrtc": lambda: TransportParams(
         audio_in_enabled=True,
         audio_out_enabled=True,
         vad_analyzer=SileroVADAnalyzer(params=VADParams(stop_secs=0.2)),
-        turn_analyzer=LocalSmartTurnAnalyzerV3(
-            smart_turn_model_path=smart_turn_model_path, params=SmartTurnParams()
-        ),
+        turn_analyzer=LocalSmartTurnAnalyzerV3(params=SmartTurnParams()),
     ),
 }
 

pyproject.toml

@@ -47,32 +47,32 @@ Website = "https://pipecat.ai"
 [project.optional-dependencies]
 aic = [ "aic-sdk~=1.0.1" ]
 anthropic = [ "anthropic~=0.49.0" ]
-assemblyai = [ "websockets>=13.1,<15.0" ]
-asyncai = [ "websockets>=13.1,<15.0" ]
-aws = [ "aioboto3~=15.0.0", "websockets>=13.1,<15.0" ]
+assemblyai = [ "pipecat-ai[websockets-base]" ]
+asyncai = [ "pipecat-ai[websockets-base]" ]
+aws = [ "aioboto3~=15.0.0", "pipecat-ai[websockets-base]" ]
 aws-nova-sonic = [ "aws_sdk_bedrock_runtime~=0.0.2; python_version>='3.12'" ]
 azure = [ "azure-cognitiveservices-speech~=1.42.0"]
-cartesia = [ "cartesia~=2.0.3", "websockets>=13.1,<15.0" ]
+cartesia = [ "cartesia~=2.0.3", "pipecat-ai[websockets-base]" ]
 cerebras = []
 deepseek = []
 daily = [ "daily-python~=0.19.9" ]
 deepgram = [ "deepgram-sdk~=4.7.0" ]
-elevenlabs = [ "websockets>=13.1,<15.0" ]
+elevenlabs = [ "pipecat-ai[websockets-base]" ]
 fal = [ "fal-client~=0.5.9" ]
 fireworks = []
-fish = [ "ormsgpack~=1.7.0", "websockets>=13.1,<15.0" ]
-gladia = [ "websockets>=13.1,<15.0" ]
-google = [ "google-cloud-speech~=2.32.0", "google-cloud-texttospeech~=2.26.0", "google-genai~=1.24.0", "websockets>=13.1,<15.0" ]
+fish = [ "ormsgpack~=1.7.0", "pipecat-ai[websockets-base]" ]
+gladia = [ "pipecat-ai[websockets-base]" ]
+google = [ "google-cloud-speech~=2.32.0", "google-cloud-texttospeech~=2.26.0", "google-genai~=1.24.0", "pipecat-ai[websockets-base]" ]
 grok = []
 groq = [ "groq~=0.23.0" ]
 gstreamer = [ "pygobject~=3.50.0" ]
-heygen = [ "livekit>=0.22.0", "websockets>=13.1,<15.0" ]
+heygen = [ "livekit>=1.0.13", "pipecat-ai[websockets-base]" ]
 inworld = []
 krisp = [ "pipecat-ai-krisp~=0.4.0" ]
 koala = [ "pvkoala~=2.0.3" ]
 langchain = [ "langchain~=0.3.20", "langchain-community~=0.3.20", "langchain-openai~=0.3.9" ]
-livekit = [ "livekit~=0.22.0", "livekit-api~=0.8.2", "tenacity>=8.2.3,<10.0.0" ]
-lmnt = [ "websockets>=13.1,<15.0" ]
+livekit = [ "livekit~=1.0.13", "livekit-api~=1.0.5", "tenacity>=8.2.3,<10.0.0" ]
+lmnt = [ "pipecat-ai[websockets-base]" ]
 local = [ "pyaudio~=0.2.14" ]
 mcp = [ "mcp[cli]~=1.9.4" ]
 mem0 = [ "mem0ai~=0.1.94" ]
@@ -80,34 +80,35 @@ mistral = []
 mlx-whisper = [ "mlx-whisper~=0.4.2" ]
 moondream = [ "accelerate~=1.10.0", "einops~=0.8.0", "pyvips[binary]~=3.0.0", "timm~=1.0.13", "transformers>=4.48.0" ]
 nim = []
-neuphonic = [ "websockets>=13.1,<15.0" ]
+neuphonic = [ "pipecat-ai[websockets-base]" ]
 noisereduce = [ "noisereduce~=3.0.3" ]
-openai = [ "websockets>=13.1,<15.0" ]
+openai = [ "pipecat-ai[websockets-base]" ]
 openpipe = [ "openpipe~=4.50.0" ]
 openrouter = []
 perplexity = []
-playht = [ "websockets>=13.1,<15.0" ]
+playht = [ "pipecat-ai[websockets-base]" ]
 qwen = []
-rime = [ "websockets>=13.1,<15.0" ]
+rime = [ "pipecat-ai[websockets-base]" ]
 riva = [ "nvidia-riva-client~=2.21.1" ]
 runner = [ "python-dotenv>=1.0.0,<2.0.0", "uvicorn>=0.32.0,<1.0.0", "fastapi>=0.115.6,<0.117.0", "pipecat-ai-small-webrtc-prebuilt>=1.0.0"]
 sambanova = []
-sarvam = [ "websockets>=13.1,<15.0" ]
+sarvam = [ "pipecat-ai[websockets-base]" ]
 sentry = [ "sentry-sdk~=2.23.1" ]
 local-smart-turn = [ "coremltools>=8.0", "transformers", "torch>=2.5.0,<3", "torchaudio>=2.5.0,<3" ]
-local-smart-turn-v3 = [ "transformers", "torch>=2.5.0,<3", "torchaudio>=2.5.0,<3", "onnxruntime>=1.20.1, <2" ]
+local-smart-turn-v3 = [ "transformers", "onnxruntime>=1.20.1, <2" ]
 remote-smart-turn = []
 silero = [ "onnxruntime>=1.20.1, <2" ]
 simli = [ "simli-ai~=0.1.10"]
-soniox = [ "websockets>=13.1,<15.0" ]
+soniox = [ "pipecat-ai[websockets-base]" ]
 soundfile = [ "soundfile~=0.13.0" ]
 speechmatics = [ "speechmatics-rt>=0.4.0" ]
 tavus=[]
 together = []
 tracing = [ "opentelemetry-sdk>=1.33.0", "opentelemetry-api>=1.33.0", "opentelemetry-instrumentation>=0.54b0" ]
 ultravox = [ "transformers>=4.48.0", "vllm>=0.9.0" ]
-webrtc = [ "aiortc~=1.11.0", "opencv-python~=4.11.0.86" ]
-websocket = [ "websockets>=13.1,<15.0", "fastapi>=0.115.6,<0.117.0" ]
+webrtc = [ "aiortc~=1.13.0", "opencv-python~=4.11.0.86" ]
+websocket = [ "pipecat-ai[websockets-base]", "fastapi>=0.115.6,<0.117.0" ]
+websockets-base = [ "websockets>=13.1,<16.0" ]
 whisper = [ "faster-whisper~=1.1.1" ]
 
 [dependency-groups]

scripts/mem-watch.sh

@@ -0,0 +1,12 @@
+#!/bin/bash
+
+PID=$1
+
+while true; do
+  # Clear the screen
+  clear
+  # Print the header + RSS in GB
+  ps -p "$PID" -o pid,comm,rss | \
+  awk 'NR==1 {print $0, "rss_GB"} NR>1 {printf "%s %s %s %.2f\n", $1,$2,$3,$3/1024/1024}'
+  sleep 1
+done

src/pipecat/audio/turn/smart_turn/local_smart_turn_v3.py

@@ -98,15 +98,15 @@ class LocalSmartTurnAnalyzerV3(BaseSmartTurn):
         inputs = self._feature_extractor(
             audio_array,
             sampling_rate=16000,
-            return_tensors="pt",
+            return_tensors="np",
             padding="max_length",
             max_length=8 * 16000,
             truncation=True,
             do_normalize=True,
         )
 
-        # Convert to numpy and ensure correct shape for ONNX
-        input_features = inputs.input_features.squeeze(0).numpy().astype(np.float32)
+        # Extract features and ensure correct shape for ONNX
+        input_features = inputs.input_features.squeeze(0).astype(np.float32)
         input_features = np.expand_dims(input_features, axis=0)  # Add batch dimension
 
         # Run ONNX inference

src/pipecat/frames/frames.py

@@ -1604,7 +1604,7 @@ class MixerEnableFrame(MixerControlFrame):
 
 @dataclass
 class ServiceSwitcherFrame(ControlFrame):
-    """A base class for frames that control ServiceSwitcher behavior."""
+    """A base class for frames that affect ServiceSwitcher behavior."""
 
     pass
 

src/pipecat/pipeline/service_switcher.py

@@ -6,9 +6,15 @@
 
 """Service switcher for switching between different services at runtime, with different switching strategies."""
 
+from dataclasses import dataclass
 from typing import Any, Generic, List, Optional, Type, TypeVar
 
-from pipecat.frames.frames import Frame, ManuallySwitchServiceFrame, ServiceSwitcherFrame
+from pipecat.frames.frames import (
+    ControlFrame,
+    Frame,
+    ManuallySwitchServiceFrame,
+    ServiceSwitcherFrame,
+)
 from pipecat.pipeline.parallel_pipeline import ParallelPipeline
 from pipecat.processors.filters.function_filter import FunctionFilter
 from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
@@ -22,19 +28,6 @@ class ServiceSwitcherStrategy:
         self.services = services
         self.active_service: Optional[FrameProcessor] = None
 
-    def is_active(self, service: FrameProcessor) -> bool:
-        """Determine if the given service is the currently active one.
-
-        This method should be overridden by subclasses to implement specific logic.
-
-        Args:
-            service: The service to check.
-
-        Returns:
-            True if the given service is the active one, False otherwise.
-        """
-        raise NotImplementedError("Subclasses must implement this method.")
-
     def handle_frame(self, frame: ServiceSwitcherFrame, direction: FrameDirection):
         """Handle a frame that controls service switching.
 
@@ -60,17 +53,6 @@ class ServiceSwitcherStrategyManual(ServiceSwitcherStrategy):
         super().__init__(services)
         self.active_service = services[0] if services else None
 
-    def is_active(self, service: FrameProcessor) -> bool:
-        """Check if the given service is the currently active one.
-
-        Args:
-            service: The service to check.
-
-        Returns:
-            True if the given service is the active one, False otherwise.
-        """
-        return service == self.active_service
-
     def handle_frame(self, frame: ServiceSwitcherFrame, direction: FrameDirection):
         """Handle a frame that controls service switching.
 
@@ -79,20 +61,21 @@ class ServiceSwitcherStrategyManual(ServiceSwitcherStrategy):
             direction: The direction of the frame (upstream or downstream).
         """
         if isinstance(frame, ManuallySwitchServiceFrame):
-            self._set_active(frame.service)
+            self._set_active_if_available(frame.service)
         else:
             raise ValueError(f"Unsupported frame type: {type(frame)}")
 
-    def _set_active(self, service: FrameProcessor):
-        """Set the active service to the given one.
+    def _set_active_if_available(self, service: FrameProcessor):
+        """Set the active service to the given one, if it is in the list of available services.
+
+        If it's not in the list, the request is ignored, as it may have been
+        intended for another ServiceSwitcher in the pipeline.
 
         Args:
             service: The service to set as active.
         """
         if service in self.services:
             self.active_service = service
-        else:
-            raise ValueError(f"Service {service} is not in the list of available services.")
 
 
 StrategyType = TypeVar("StrategyType", bound=ServiceSwitcherStrategy)
@@ -108,6 +91,43 @@ class ServiceSwitcher(ParallelPipeline, Generic[StrategyType]):
         self.services = services
         self.strategy = strategy
 
+    class ServiceSwitcherFilter(FunctionFilter):
+        """An internal filter that allows frames to pass through to the wrapped service only if it's the active service."""
+
+        def __init__(
+            self,
+            wrapped_service: FrameProcessor,
+            active_service: FrameProcessor,
+            direction: FrameDirection,
+        ):
+            """Initialize the service switcher filter with a strategy and direction."""
+
+            async def filter(_: Frame) -> bool:
+                return self._wrapped_service == self._active_service
+
+            super().__init__(filter, direction)
+            self._wrapped_service = wrapped_service
+            self._active_service = active_service
+
+        async def process_frame(self, frame, direction):
+            """Process a frame through the filter, handling special internal filter-updating frames."""
+            if isinstance(frame, ServiceSwitcher.ServiceSwitcherFilterFrame):
+                self._active_service = frame.active_service
+                # Two ServiceSwitcherFilters "sandwich" a service. Push the
+                # frame only to update the other side of the sandwich, but
+                # otherwise don't let it leave the sandwich.
+                if direction == self._direction:
+                    await self.push_frame(frame, direction)
+                return
+
+            await super().process_frame(frame, direction)
+
+    @dataclass
+    class ServiceSwitcherFilterFrame(ControlFrame):
+        """An internal frame used by ServiceSwitcher to filter frames based on active service."""
+
+        active_service: FrameProcessor
+
     @staticmethod
     def _make_pipeline_definitions(
         services: List[FrameProcessor], strategy: ServiceSwitcherStrategy
@@ -121,14 +141,18 @@ class ServiceSwitcher(ParallelPipeline, Generic[StrategyType]):
     def _make_pipeline_definition(
         service: FrameProcessor, strategy: ServiceSwitcherStrategy
     ) -> Any:
-        async def filter(frame) -> bool:
-            _ = frame
-            return strategy.is_active(service)
-
         return [
-            FunctionFilter(filter, direction=FrameDirection.DOWNSTREAM),
+            ServiceSwitcher.ServiceSwitcherFilter(
+                wrapped_service=service,
+                active_service=strategy.active_service,
+                direction=FrameDirection.DOWNSTREAM,
+            ),
             service,
-            FunctionFilter(filter, direction=FrameDirection.UPSTREAM),
+            ServiceSwitcher.ServiceSwitcherFilter(
+                wrapped_service=service,
+                active_service=strategy.active_service,
+                direction=FrameDirection.UPSTREAM,
+            ),
         ]
 
     async def process_frame(self, frame: Frame, direction: FrameDirection):
@@ -142,3 +166,7 @@ class ServiceSwitcher(ParallelPipeline, Generic[StrategyType]):
 
         if isinstance(frame, ServiceSwitcherFrame):
             self.strategy.handle_frame(frame, direction)
+            service_switcher_filter_frame = ServiceSwitcher.ServiceSwitcherFilterFrame(
+                active_service=self.strategy.active_service
+            )
+            await super().process_frame(service_switcher_filter_frame, direction)

src/pipecat/processors/frame_processor.py

@@ -220,6 +220,11 @@ class FrameProcessor(BaseObject):
         self.__process_event: Optional[asyncio.Event] = None
         self.__process_frame_task: Optional[asyncio.Task] = None
 
+        # To interrupt a pipeline, we push an `InterruptionTaskFrame` upstream.
+        # Then we wait for the corresponding `InterruptionFrame` to travel from
+        # the start of the pipeline back to the processor that sent the
+        # `InterruptionTaskFrame`. This wait is handled using the following
+        # event.
         self._wait_for_interruption = False
         self._wait_interruption_event = asyncio.Event()
 
@@ -632,7 +637,9 @@ class FrameProcessor(BaseObject):
 
         await self.__internal_push_frame(frame, direction)
 
-        if isinstance(frame, InterruptionFrame):
+        # If we are waiting for an interruption and we get an interruption, then
+        # we can unblock `push_interruption_task_frame_and_wait()`.
+        if self._wait_for_interruption and isinstance(frame, InterruptionFrame):
             self._wait_interruption_event.set()
 
     async def push_interruption_task_frame_and_wait(self):

src/pipecat/processors/user_idle_processor.py

@@ -17,7 +17,6 @@ from pipecat.frames.frames import (
     Frame,
     FunctionCallInProgressFrame,
     FunctionCallResultFrame,
-    StartFrame,
     UserStartedSpeakingFrame,
     UserStoppedSpeakingFrame,
 )
@@ -185,15 +184,13 @@ class UserIdleProcessor(FrameProcessor):
 
         Runs in a loop until cancelled or callback indicates completion.
         """
-        while True:
+        running = True
+        while running:
             try:
                 await asyncio.wait_for(self._idle_event.wait(), timeout=self._timeout)
             except asyncio.TimeoutError:
                 if not self._interrupted:
                     self._retry_count += 1
-                    should_continue = await self._callback(self, self._retry_count)
-                    if not should_continue:
-                        await self._stop()
-                        break
+                    running = await self._callback(self, self._retry_count)
             finally:
                 self._idle_event.clear()

src/pipecat/runner/run.py

@@ -70,7 +70,6 @@ import asyncio
 import os
 import sys
 from contextlib import asynccontextmanager
-from typing import Dict
 
 from loguru import logger
 
@@ -79,6 +78,10 @@ from pipecat.runner.types import (
     SmallWebRTCRunnerArguments,
     WebSocketRunnerArguments,
 )
+from pipecat.transports.smallwebrtc.request_handler import (
+    SmallWebRTCRequest,
+    SmallWebRTCRequestHandler,
+)
 
 try:
     import uvicorn
@@ -187,9 +190,6 @@ def _setup_webrtc_routes(app: FastAPI, esp32_mode: bool = False, host: str = "lo
         logger.error(f"WebRTC transport dependencies not installed: {e}")
         return
 
-    # Store connections by pc_id
-    pcs_map: Dict[str, SmallWebRTCConnection] = {}
-
     # Mount the frontend
     app.mount("/client", SmallWebRTCPrebuiltUI)
 
@@ -198,51 +198,33 @@ def _setup_webrtc_routes(app: FastAPI, esp32_mode: bool = False, host: str = "lo
         """Redirect root requests to client interface."""
         return RedirectResponse(url="/client/")
 
+    # Initialize the SmallWebRTC request handler
+    small_webrtc_handler: SmallWebRTCRequestHandler = SmallWebRTCRequestHandler(
+        esp32_mode=esp32_mode, host=host
+    )
+
     @app.post("/api/offer")
-    async def offer(request: dict, background_tasks: BackgroundTasks):
-        """Handle WebRTC offer requests and manage peer connections."""
-        pc_id = request.get("pc_id")
-
-        if pc_id and pc_id in pcs_map:
-            pipecat_connection = pcs_map[pc_id]
-            logger.info(f"Reusing existing connection for pc_id: {pc_id}")
-            await pipecat_connection.renegotiate(
-                sdp=request["sdp"],
-                type=request["type"],
-                restart_pc=request.get("restart_pc", False),
-            )
-        else:
-            pipecat_connection = SmallWebRTCConnection()
-            await pipecat_connection.initialize(sdp=request["sdp"], type=request["type"])
-
-            @pipecat_connection.event_handler("closed")
-            async def handle_disconnected(webrtc_connection: SmallWebRTCConnection):
-                """Handle WebRTC connection closure and cleanup."""
-                logger.info(f"Discarding peer connection for pc_id: {webrtc_connection.pc_id}")
-                pcs_map.pop(webrtc_connection.pc_id, None)
+    async def offer(request: SmallWebRTCRequest, background_tasks: BackgroundTasks):
+        """Handle WebRTC offer requests via SmallWebRTCRequestHandler."""
 
+        # Prepare runner arguments with the callback to run your bot
+        async def webrtc_connection_callback(connection):
             bot_module = _get_bot_module()
-            runner_args = SmallWebRTCRunnerArguments(webrtc_connection=pipecat_connection)
+            runner_args = SmallWebRTCRunnerArguments(webrtc_connection=connection)
             background_tasks.add_task(bot_module.bot, runner_args)
 
-        answer = pipecat_connection.get_answer()
-
-        # Apply ESP32 SDP munging if enabled
-        if esp32_mode and host != "localhost":
-            from pipecat.runner.utils import smallwebrtc_sdp_munging
-
-            answer["sdp"] = smallwebrtc_sdp_munging(answer["sdp"], host)
-
-        pcs_map[answer["pc_id"]] = pipecat_connection
+        # Delegate handling to SmallWebRTCRequestHandler
+        answer = await small_webrtc_handler.handle_web_request(
+            request=request,
+            webrtc_connection_callback=webrtc_connection_callback,
+        )
         return answer
 
     @asynccontextmanager
     async def lifespan(app: FastAPI):
         """Manage FastAPI application lifecycle and cleanup connections."""
         yield
-        coros = [pc.disconnect() for pc in pcs_map.values()]
-        await asyncio.gather(*coros)
-        pcs_map.clear()
+        await small_webrtc_handler.close()
 
     app.router.lifespan_context = lifespan
 

src/pipecat/services/asyncai/tts.py

@@ -119,7 +119,6 @@ class AsyncAITTSService(InterruptibleTTSService):
         """
         super().__init__(
             aggregate_sentences=aggregate_sentences,
-            push_text_frames=False,
             pause_frame_processing=True,
             push_stop_frames=True,
             sample_rate=sample_rate,

src/pipecat/services/aws/stt.py

@@ -532,9 +532,7 @@ class AWSTranscribeSTTService(STTService):
                     logger.debug(f"{self} Other message type received: {headers}")
                     logger.debug(f"{self} Payload: {payload}")
             except websockets.exceptions.ConnectionClosed as e:
-                logger.error(
-                    f"{self} WebSocket connection closed in receive loop with code {e.code}: {e.reason}"
-                )
+                logger.error(f"{self} WebSocket connection closed in receive loop: {e}")
                 break
             except Exception as e:
                 logger.error(f"{self} Unexpected error in receive loop: {e}")

src/pipecat/services/gladia/stt.py

@@ -13,6 +13,7 @@ supporting multiple languages, custom vocabulary, and various audio processing o
 import asyncio
 import base64
 import json
+import warnings
 from typing import Any, AsyncGenerator, Dict, Literal, Optional
 
 import aiohttp
@@ -173,8 +174,6 @@ class _InputParamsDescriptor:
     """Descriptor for backward compatibility with deprecation warning."""
 
     def __get__(self, obj, objtype=None):
-        import warnings
-
         with warnings.catch_warnings():
             warnings.simplefilter("always")
             warnings.warn(
@@ -208,7 +207,7 @@ class GladiaSTTService(STTService):
         api_key: str,
         region: Literal["us-west", "eu-west"] | None = None,
         url: str = "https://api.gladia.io/v2/live",
-        confidence: float = 0.5,
+        confidence: Optional[float] = None,
         sample_rate: Optional[int] = None,
         model: str = "solaria-1",
         params: Optional[GladiaInputParams] = None,
@@ -224,6 +223,11 @@ class GladiaSTTService(STTService):
             region: Region used to process audio. eu-west or us-west. Defaults to eu-west.
             url: Gladia API URL. Defaults to "https://api.gladia.io/v2/live".
             confidence: Minimum confidence threshold for transcriptions (0.0-1.0).
+
+                .. deprecated:: 0.0.86
+                    The 'confidence' parameter is deprecated and will be removed in a future version.
+                    No confidence threshold is applied.
+
             sample_rate: Audio sample rate in Hz. If None, uses service default.
             model: Model to use for transcription. Defaults to "solaria-1".
             params: Additional configuration parameters for Gladia service.
@@ -236,7 +240,6 @@ class GladiaSTTService(STTService):
 
         params = params or GladiaInputParams()
 
-        # Warn about deprecated language parameter if it's used
         if params.language is not None:
             with warnings.catch_warnings():
                 warnings.simplefilter("always")
@@ -247,11 +250,20 @@ class GladiaSTTService(STTService):
                     stacklevel=2,
                 )
 
+        if confidence:
+            with warnings.catch_warnings():
+                warnings.simplefilter("always")
+                warnings.warn(
+                    "The 'confidence' parameter is deprecated and will be removed in a future version. "
+                    "No confidence threshold is applied.",
+                    DeprecationWarning,
+                    stacklevel=2,
+                )
+
         self._api_key = api_key
         self._region = region
         self._url = url
         self.set_model_name(model)
-        self._confidence = confidence
         self._params = params
         self._websocket = None
         self._receive_task = None
@@ -575,43 +587,40 @@ class GladiaSTTService(STTService):
 
                 elif content["type"] == "transcript":
                     utterance = content["data"]["utterance"]
-                    confidence = utterance.get("confidence", 0)
                     language = utterance["language"]
                     transcript = utterance["text"]
                     is_final = content["data"]["is_final"]
-                    if confidence >= self._confidence:
-                        if is_final:
-                            await self.push_frame(
-                                TranscriptionFrame(
-                                    transcript,
-                                    self._user_id,
-                                    time_now_iso8601(),
-                                    language,
-                                    result=content,
-                                )
+                    if is_final:
+                        await self.push_frame(
+                            TranscriptionFrame(
+                                transcript,
+                                self._user_id,
+                                time_now_iso8601(),
+                                language,
+                                result=content,
                             )
-                            await self._handle_transcription(
-                                transcript=transcript,
-                                is_final=is_final,
-                                language=language,
-                            )
-                        else:
-                            await self.push_frame(
-                                InterimTranscriptionFrame(
-                                    transcript,
-                                    self._user_id,
-                                    time_now_iso8601(),
-                                    language,
-                                    result=content,
-                                )
+                        )
+                        await self._handle_transcription(
+                            transcript=transcript,
+                            is_final=is_final,
+                            language=language,
+                        )
+                    else:
+                        await self.push_frame(
+                            InterimTranscriptionFrame(
+                                transcript,
+                                self._user_id,
+                                time_now_iso8601(),
+                                language,
+                                result=content,
                             )
+                        )
                 elif content["type"] == "translation":
                     translated_utterance = content["data"]["translated_utterance"]
                     original_language = content["data"]["original_language"]
                     translated_language = translated_utterance["language"]
-                    confidence = translated_utterance.get("confidence", 0)
                     translation = translated_utterance["text"]
-                    if translated_language != original_language and confidence >= self._confidence:
+                    if translated_language != original_language:
                         await self.push_frame(
                             TranslationFrame(
                                 translation, "", time_now_iso8601(), translated_language

src/pipecat/services/google/llm_vertex.py

@@ -83,14 +83,23 @@ class GoogleVertexLLMService(OpenAILLMService):
         self._api_key = self._get_api_token(credentials, credentials_path)
 
         super().__init__(
-            api_key=self._api_key, base_url=base_url, model=model, params=params, **kwargs
+            api_key=self._api_key,
+            base_url=base_url,
+            model=model,
+            params=params,
+            **kwargs,
         )
 
     @staticmethod
     def _get_base_url(params: InputParams) -> str:
         """Construct the base URL for Vertex AI API."""
+        # Determine the correct API host based on location
+        if params.location == "global":
+            api_host = "aiplatform.googleapis.com"
+        else:
+            api_host = f"{params.location}-aiplatform.googleapis.com"
         return (
-            f"https://{params.location}-aiplatform.googleapis.com/v1/"
+            f"https://{api_host}/v1/"
             f"projects/{params.project_id}/locations/{params.location}/endpoints/openapi"
         )
 
@@ -118,12 +127,14 @@ class GoogleVertexLLMService(OpenAILLMService):
         if credentials:
             # Parse and load credentials from JSON string
             creds = service_account.Credentials.from_service_account_info(
-                json.loads(credentials), scopes=["https://www.googleapis.com/auth/cloud-platform"]
+                json.loads(credentials),
+                scopes=["https://www.googleapis.com/auth/cloud-platform"],
             )
         elif credentials_path:
             # Load credentials from JSON file
             creds = service_account.Credentials.from_service_account_file(
-                credentials_path, scopes=["https://www.googleapis.com/auth/cloud-platform"]
+                credentials_path,
+                scopes=["https://www.googleapis.com/auth/cloud-platform"],
             )
         else:
             try:

src/pipecat/services/mcp_service.py

@@ -7,7 +7,7 @@
 """MCP (Model Context Protocol) client for integrating external tools with LLMs."""
 
 import json
-from typing import Any, Dict, List, Tuple
+from typing import Any, Dict, List, TypeAlias
 
 from loguru import logger
 
@@ -28,6 +28,8 @@ except ModuleNotFoundError as e:
     logger.error("In order to use an MCP client, you need to `pip install pipecat-ai[mcp]`.")
     raise Exception(f"Missing module: {e}")
 
+ServerParameters: TypeAlias = StdioServerParameters | SseServerParameters | StreamableHttpParameters
+
 
 class MCPClient(BaseObject):
     """Client for Model Context Protocol (MCP) servers.
@@ -42,7 +44,7 @@ class MCPClient(BaseObject):
 
     def __init__(
         self,
-        server_params: Tuple[StdioServerParameters, SseServerParameters, StreamableHttpParameters],
+        server_params: ServerParameters,
         **kwargs,
     ):
         """Initialize the MCP client with server parameters.

src/pipecat/transports/daily/transport.py

@@ -359,6 +359,7 @@ class DailyTransportClient(EventHandler):
         self._transcription_ids = []
         self._transcription_status = None
         self._dial_out_session_id: str = ""
+        self._dial_in_session_id: str = ""
 
         self._joining = False
         self._joined = False
@@ -823,6 +824,16 @@ class DailyTransportClient(EventHandler):
         Args:
             settings: SIP call transfer settings.
         """
+        session_id = (
+            settings.get("sessionId") or self._dial_out_session_id or self._dial_in_session_id
+        )
+        if not session_id:
+            logger.error("Unable to transfer SIP call: 'sessionId' is not set")
+            return
+
+        # Update 'sessionId' field.
+        settings["sessionId"] = session_id
+
         future = self._get_event_loop().create_future()
         self._client.sip_call_transfer(settings, completion=completion_callback(future))
         await future
@@ -1141,6 +1152,7 @@ class DailyTransportClient(EventHandler):
         Args:
             data: Dial-in connection data.
         """
+        self._dial_in_session_id = data["sessionId"] if "sessionId" in data else ""
         self._call_event_callback(self._callbacks.on_dialin_connected, data)
 
     def on_dialin_ready(self, sip_endpoint: str):
@@ -1157,6 +1169,9 @@ class DailyTransportClient(EventHandler):
         Args:
             data: Dial-in stop data.
         """
+        # Cleanup only if our session stopped.
+        if data.get("sessionId") == self._dial_in_session_id:
+            self._dial_in_session_id = ""
         self._call_event_callback(self._callbacks.on_dialin_stopped, data)
 
     def on_dialin_error(self, data: Any):
@@ -1165,6 +1180,9 @@ class DailyTransportClient(EventHandler):
         Args:
             data: Dial-in error data.
         """
+        # Cleanup only if our session errored out.
+        if data.get("sessionId") == self._dial_in_session_id:
+            self._dial_in_session_id = ""
         self._call_event_callback(self._callbacks.on_dialin_error, data)
 
     def on_dialin_warning(self, data: Any):
@@ -1199,7 +1217,7 @@ class DailyTransportClient(EventHandler):
             data: Dial-out stop data.
         """
         # Cleanup only if our session stopped.
-        if data["sessionId"] == self._dial_out_session_id:
+        if data.get("sessionId") == self._dial_out_session_id:
             self._dial_out_session_id = ""
         self._call_event_callback(self._callbacks.on_dialout_stopped, data)
 
@@ -1210,7 +1228,7 @@ class DailyTransportClient(EventHandler):
             data: Dial-out error data.
         """
         # Cleanup only if our session errored out.
-        if data["sessionId"] == self._dial_out_session_id:
+        if data.get("sessionId") == self._dial_out_session_id:
             self._dial_out_session_id = ""
         self._call_event_callback(self._callbacks.on_dialout_error, data)
 

src/pipecat/transports/smallwebrtc/connection.py

@@ -95,15 +95,20 @@ class SmallWebRTCTrack:
     enable/disable control and frame discarding for audio and video streams.
     """
 
-    def __init__(self, track: MediaStreamTrack):
+    def __init__(self, receiver):
         """Initialize the WebRTC track wrapper.
 
         Args:
-            track: The underlying MediaStreamTrack to wrap.
-            index: The index of the track in the transceiver (0 for mic, 1 for cam, 2 for screen)
+            receiver: The RemoteStreamTrack receiver instance.
         """
-        self._track = track
+        self._receiver = receiver
+        # Configuring the receiver for not consuming the track by default to prevent memory grow
+        self._receiver._enabled = False
+        self._track = receiver.track
         self._enabled = True
+        self._last_recv_time: float = 0.0
+        self._idle_task: Optional[asyncio.Task] = None
+        self._idle_timeout: float = 2.0  # seconds before discarding old frames
 
     def set_enabled(self, enabled: bool) -> None:
         """Enable or disable the track.
@@ -138,13 +143,44 @@ class SmallWebRTCTrack:
     async def recv(self) -> Optional[Frame]:
         """Receive the next frame from the track.
 
+        Enables the internal receiving state and starts idle watcher.
+
         Returns:
             The next frame, except for video tracks, where it returns the frame only if the track is enabled, otherwise, returns None.
         """
+        self._receiver._enabled = True
+        self._last_recv_time = time.time()
+
+        # start idle watcher if not already running
+        if not self._idle_task or self._idle_task.done():
+            self._idle_task = asyncio.create_task(self._idle_watcher())
+
         if not self._enabled and self._track.kind == "video":
             return None
         return await self._track.recv()
 
+    async def _idle_watcher(self):
+        """Disable receiving if idle for more than _idle_timeout and monitor queue size."""
+        while self._receiver._enabled:
+            await asyncio.sleep(self._idle_timeout)
+            idle_duration = time.time() - self._last_recv_time
+            if idle_duration >= self._idle_timeout:
+                # discard old frames to prevent memory growth
+                logger.debug(
+                    f"Disabling receiver for {self._track.kind} track after {idle_duration:.2f}s idle"
+                )
+                await self.discard_old_frames()
+                self._receiver._enabled = False
+
+    def stop(self):
+        """Stop receiving frames from the track."""
+        self._receiver._enabled = False
+        if self._idle_task:
+            self._idle_task.cancel()
+            self._idle_task = None
+        if self._track:
+            self._track.stop()
+
     def __getattr__(self, name):
         """Forward attribute access to the underlying track.
 
@@ -454,6 +490,10 @@ class SmallWebRTCConnection(BaseObject):
 
     async def _close(self):
         """Close the peer connection and cleanup resources."""
+        for track in self._track_map.values():
+            if track:
+                track.stop()
+        self._track_map.clear()
         if self._pc:
             await self._pc.close()
         self._message_queue.clear()
@@ -526,8 +566,8 @@ class SmallWebRTCConnection(BaseObject):
             logger.warning("No audio transceiver is available")
             return None
 
-        track = transceivers[AUDIO_TRANSCEIVER_INDEX].receiver.track
-        audio_track = SmallWebRTCTrack(track) if track else None
+        receiver = transceivers[AUDIO_TRANSCEIVER_INDEX].receiver
+        audio_track = SmallWebRTCTrack(receiver) if receiver else None
         self._track_map[AUDIO_TRANSCEIVER_INDEX] = audio_track
         return audio_track
 
@@ -548,8 +588,8 @@ class SmallWebRTCConnection(BaseObject):
             logger.warning("No video transceiver is available")
             return None
 
-        track = transceivers[VIDEO_TRANSCEIVER_INDEX].receiver.track
-        video_track = SmallWebRTCTrack(track) if track else None
+        receiver = transceivers[VIDEO_TRANSCEIVER_INDEX].receiver
+        video_track = SmallWebRTCTrack(receiver) if receiver else None
         self._track_map[VIDEO_TRANSCEIVER_INDEX] = video_track
         return video_track
 
@@ -570,8 +610,8 @@ class SmallWebRTCConnection(BaseObject):
             logger.warning("No screen video transceiver is available")
             return None
 
-        track = transceivers[SCREEN_VIDEO_TRANSCEIVER_INDEX].receiver.track
-        video_track = SmallWebRTCTrack(track) if track else None
+        receiver = transceivers[SCREEN_VIDEO_TRANSCEIVER_INDEX].receiver
+        video_track = SmallWebRTCTrack(receiver) if receiver else None
         self._track_map[SCREEN_VIDEO_TRANSCEIVER_INDEX] = video_track
         return video_track
 

src/pipecat/transports/smallwebrtc/request_handler.py

@@ -0,0 +1,200 @@
+#
+# Copyright (c) 2024–2025, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""SmallWebRTC request handler for managing peer connections.
+
+This module provides a client for handling web requests and managing WebRTC connections.
+"""
+
+import asyncio
+from dataclasses import dataclass
+from enum import Enum
+from typing import Any, Awaitable, Callable, Dict, List, Optional
+
+from fastapi import HTTPException
+from loguru import logger
+
+from pipecat.transports.smallwebrtc.connection import IceServer, SmallWebRTCConnection
+
+
+@dataclass
+class SmallWebRTCRequest:
+    """Small WebRTC transport session arguments for the runner.
+
+    Parameters:
+        sdp: The SDP string (Session Description Protocol).
+        type: The type of the SDP, either "offer" or "answer".
+        pc_id: Optional identifier for the peer connection.
+        restart_pc: Optional whether to restart the peer connection.
+        request_data: Optional custom data sent by the customer.
+    """
+
+    sdp: str
+    type: str
+    pc_id: Optional[str] = None
+    restart_pc: Optional[bool] = None
+    request_data: Optional[Any] = None
+
+
+class ConnectionMode(Enum):
+    """Enum defining the connection handling modes."""
+
+    SINGLE = "single"  # Only one active connection allowed
+    MULTIPLE = "multiple"  # Multiple simultaneous connections allowed
+
+
+class SmallWebRTCRequestHandler:
+    """SmallWebRTC request handler for managing peer connections.
+
+    This class is responsible for:
+      - Handling incoming SmallWebRTC requests.
+      - Creating and managing WebRTC peer connections.
+      - Supporting ESP32-specific SDP munging if enabled.
+      - Invoking callbacks for newly initialized connections.
+      - Supporting both single and multiple connection modes.
+    """
+
+    def __init__(
+        self,
+        ice_servers: Optional[List[IceServer]] = None,
+        esp32_mode: bool = False,
+        host: Optional[str] = None,
+        connection_mode: ConnectionMode = ConnectionMode.MULTIPLE,
+    ) -> None:
+        """Initialize a SmallWebRTC request handler.
+
+        Args:
+            ice_servers (Optional[List[IceServer]]): List of ICE servers to use for WebRTC
+                connections.
+            esp32_mode (bool): If True, enables ESP32-specific SDP munging.
+            host (Optional[str]): Host address used for SDP munging in ESP32 mode.
+                Ignored if `esp32_mode` is False.
+            connection_mode (ConnectionMode): Mode of operation for handling connections.
+                SINGLE allows only one active connection, MULTIPLE allows several.
+        """
+        self._ice_servers = ice_servers
+        self._esp32_mode = esp32_mode
+        self._host = host
+        self._connection_mode = connection_mode
+
+        # Store connections by pc_id
+        self._pcs_map: Dict[str, SmallWebRTCConnection] = {}
+
+    def _check_single_connection_constraints(self, pc_id: Optional[str]) -> None:
+        """Check if the connection request satisfies single connection mode constraints.
+
+        Args:
+            pc_id: The peer connection ID from the request
+
+        Raises:
+            HTTPException: If constraints are violated in single connection mode
+        """
+        if self._connection_mode != ConnectionMode.SINGLE:
+            return
+
+        if not self._pcs_map:  # No existing connections
+            return
+
+        # Get the existing connection (should be only one in single mode)
+        existing_connection = next(iter(self._pcs_map.values()))
+
+        if existing_connection.pc_id != pc_id and pc_id:
+            logger.warning(
+                f"Connection pc_id mismatch: existing={existing_connection.pc_id}, received={pc_id}"
+            )
+            raise HTTPException(status_code=400, detail="PC ID mismatch with existing connection")
+
+        if not pc_id:
+            logger.warning(
+                "Cannot create new connection: existing connection found but no pc_id received"
+            )
+            raise HTTPException(
+                status_code=400,
+                detail="Cannot create new connection with existing connection active",
+            )
+
+    async def handle_web_request(
+        self,
+        request: SmallWebRTCRequest,
+        webrtc_connection_callback: Callable[[Any], Awaitable[None]],
+    ) -> None:
+        """Handle a SmallWebRTC request and resolve the pending answer.
+
+        This method will:
+          - Reuse an existing WebRTC connection if `pc_id` exists.
+          - Otherwise, create a new `SmallWebRTCConnection`.
+          - Invoke the provided callback with the connection.
+          - Manage ESP32-specific munging if enabled.
+          - Enforce single/multiple connection mode constraints.
+
+        Args:
+            request (SmallWebRTCRequest): The incoming WebRTC request, containing
+                SDP, type, and optionally a `pc_id`.
+            webrtc_connection_callback (Callable[[Any], Awaitable[None]]): An
+                asynchronous callback function that is invoked with the WebRTC connection.
+
+        Raises:
+            HTTPException: If connection mode constraints are violated
+            Exception: Any exception raised during request handling or callback execution
+                will be logged and propagated.
+        """
+        try:
+            pc_id = request.pc_id
+
+            # Check connection mode constraints first
+            self._check_single_connection_constraints(pc_id)
+
+            # After constraints are satisfied, get the existing connection if any
+            existing_connection = self._pcs_map.get(pc_id) if pc_id else None
+
+            if existing_connection:
+                pipecat_connection = existing_connection
+                logger.info(f"Reusing existing connection for pc_id: {pc_id}")
+                await pipecat_connection.renegotiate(
+                    sdp=request.sdp,
+                    type=request.type,
+                    restart_pc=request.restart_pc or False,
+                )
+            else:
+                pipecat_connection = SmallWebRTCConnection(ice_servers=self._ice_servers)
+                await pipecat_connection.initialize(sdp=request.sdp, type=request.type)
+
+                @pipecat_connection.event_handler("closed")
+                async def handle_disconnected(webrtc_connection: SmallWebRTCConnection):
+                    logger.info(f"Discarding peer connection for pc_id: {webrtc_connection.pc_id}")
+                    self._pcs_map.pop(webrtc_connection.pc_id, None)
+
+                # Invoke callback provided in runner arguments
+                try:
+                    await webrtc_connection_callback(pipecat_connection)
+                    logger.debug(
+                        f"webrtc_connection_callback executed successfully for peer: {pipecat_connection.pc_id}"
+                    )
+                except Exception as callback_error:
+                    logger.error(
+                        f"webrtc_connection_callback failed for peer {pipecat_connection.pc_id}: {callback_error}"
+                    )
+
+            answer = pipecat_connection.get_answer()
+
+            if self._esp32_mode and self._host and self._host != "localhost":
+                from pipecat.runner.utils import smallwebrtc_sdp_munging
+
+                answer["sdp"] = smallwebrtc_sdp_munging(answer["sdp"], self._host)
+
+            self._pcs_map[answer["pc_id"]] = pipecat_connection
+
+            return answer
+        except Exception as e:
+            logger.error(f"Error processing SmallWebRTC request: {e}")
+            logger.debug(f"SmallWebRTC request details: {request}")
+            raise
+
+    async def close(self):
+        """Clear the connection map."""
+        coros = [pc.disconnect() for pc in self._pcs_map.values()]
+        await asyncio.gather(*coros)
+        self._pcs_map.clear()

tests/test_frame_processor.py

@@ -0,0 +1,67 @@
+#
+# Copyright (c) 2024-2025 Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import unittest
+
+from pipecat.frames.frames import (
+    EndFrame,
+    Frame,
+    InterruptionFrame,
+    TextFrame,
+    TransportMessageUrgentFrame,
+)
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.tests.utils import SleepFrame, run_test
+
+
+class TestFrameProcessor(unittest.IsolatedAsyncioTestCase):
+    async def test_interruption_and_wait(self):
+        class DelayFrameProcessor(FrameProcessor):
+            """This processors just gives time to the event loop to change
+            between tasks. Otherwise things happen to fast."""
+
+            async def process_frame(self, frame: Frame, direction: FrameDirection):
+                await super().process_frame(frame, direction)
+                await asyncio.sleep(0.1)
+                await self.push_frame(frame, direction)
+
+        class InterruptFrameProcessor(FrameProcessor):
+            async def process_frame(self, frame: Frame, direction: FrameDirection):
+                await super().process_frame(frame, direction)
+
+                if isinstance(frame, TextFrame):
+                    await self.push_interruption_task_frame_and_wait()
+                    await self.push_frame(TransportMessageUrgentFrame(message=frame.text))
+                else:
+                    await self.push_frame(frame, direction)
+
+        pipeline = Pipeline([DelayFrameProcessor(), InterruptFrameProcessor()])
+
+        frames_to_send = [
+            # Just a random interruption to make sure we don't clear anything
+            # before the actual `InterruptionTaskFrame` interruption.
+            InterruptionFrame(),
+            # This will generate an `InterruptionTaskFrame` and will wait for an
+            # `InterruptionFrame`.
+            TextFrame(text="Hello from Pipecat!"),
+            # Just give time for everything to complete.
+            SleepFrame(sleep=0.5),
+            EndFrame(),
+        ]
+        expected_down_frames = [
+            InterruptionFrame,
+            InterruptionFrame,
+            TransportMessageUrgentFrame,
+            EndFrame,
+        ]
+        await run_test(
+            pipeline,
+            frames_to_send=frames_to_send,
+            expected_down_frames=expected_down_frames,
+            send_end_frame=False,
+        )

tests/test_service_switcher.py

@@ -0,0 +1,303 @@
+#
+# Copyright (c) 2025, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Unit tests for ServiceSwitcher and related components."""
+
+import unittest
+
+from pipecat.frames.frames import (
+    Frame,
+    ManuallySwitchServiceFrame,
+    TextFrame,
+)
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.service_switcher import ServiceSwitcher, ServiceSwitcherStrategyManual
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.tests.utils import run_test
+
+
+class MockFrameProcessor(FrameProcessor):
+    """A test frame processor that tracks which frames it has processed."""
+
+    def __init__(self, test_name: str, **kwargs):
+        """Initialize the test processor with a name.
+
+        Args:
+            test_name: A unique name for this processor instance.
+            **kwargs: Additional arguments passed to the parent FrameProcessor.
+        """
+        super().__init__(name=test_name, **kwargs)
+        self.test_name = test_name
+        self.processed_frames = []
+        self.frame_count = 0
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        """Process an incoming frame and track it.
+
+        Args:
+            frame: The frame to process.
+            direction: The direction of frame flow in the pipeline.
+        """
+        await super().process_frame(frame, direction)
+        self.processed_frames.append(frame)
+        self.frame_count += 1
+        await self.push_frame(frame, direction)
+
+    def reset_counters(self):
+        """Reset the frame tracking counters."""
+        self.processed_frames = []
+        self.frame_count = 0
+
+
+class TestServiceSwitcherStrategyManual(unittest.IsolatedAsyncioTestCase):
+    """Test cases for ServiceSwitcherStrategyManual."""
+
+    def setUp(self):
+        """Set up test fixtures."""
+        self.service1 = MockFrameProcessor("service1")
+        self.service2 = MockFrameProcessor("service2")
+        self.service3 = MockFrameProcessor("service3")
+        self.services = [self.service1, self.service2, self.service3]
+
+    def test_init_with_services(self):
+        """Test initialization with a list of services."""
+        strategy = ServiceSwitcherStrategyManual(self.services)
+
+        self.assertEqual(strategy.services, self.services)
+        self.assertEqual(strategy.active_service, self.service1)  # First service should be active
+
+    def test_init_with_empty_services(self):
+        """Test initialization with an empty list of services."""
+        strategy = ServiceSwitcherStrategyManual([])
+
+        self.assertEqual(strategy.services, [])
+        self.assertIsNone(strategy.active_service)
+
+    def test_handle_manually_switch_service_frame(self):
+        """Test manual service switching with ManuallySwitchServiceFrame."""
+        strategy = ServiceSwitcherStrategyManual(self.services)
+
+        # Initially service1 should be active
+        self.assertEqual(strategy.active_service, self.service1)
+        self.assertNotEqual(strategy.active_service, self.service2)
+
+        # Switch to service2
+        switch_frame = ManuallySwitchServiceFrame(service=self.service2)
+        strategy.handle_frame(switch_frame, FrameDirection.DOWNSTREAM)
+
+        self.assertNotEqual(strategy.active_service, self.service1)
+        self.assertEqual(strategy.active_service, self.service2)
+        self.assertNotEqual(strategy.active_service, self.service3)
+
+        # Switch to service3
+        switch_frame = ManuallySwitchServiceFrame(service=self.service3)
+        strategy.handle_frame(switch_frame, FrameDirection.DOWNSTREAM)
+
+        self.assertNotEqual(strategy.active_service, self.service1)
+        self.assertNotEqual(strategy.active_service, self.service2)
+        self.assertEqual(strategy.active_service, self.service3)
+
+    def test_handle_frame_unsupported_frame_type(self):
+        """Test that unsupported frame types raise an error."""
+        strategy = ServiceSwitcherStrategyManual(self.services)
+        unsupported_frame = TextFrame(text="test")  # Not a ServiceSwitcherFrame
+
+        with self.assertRaises(ValueError) as context:
+            strategy.handle_frame(unsupported_frame, FrameDirection.DOWNSTREAM)
+
+        self.assertIn("Unsupported frame type", str(context.exception))
+
+
+class TestServiceSwitcher(unittest.IsolatedAsyncioTestCase):
+    """Test cases for ServiceSwitcher."""
+
+    def setUp(self):
+        """Set up test fixtures."""
+        self.service1 = MockFrameProcessor("service1")
+        self.service2 = MockFrameProcessor("service2")
+        self.service3 = MockFrameProcessor("service3")
+        self.services = [self.service1, self.service2, self.service3]
+
+    def test_init_with_manual_strategy(self):
+        """Test initialization with manual strategy."""
+        switcher = ServiceSwitcher(self.services, ServiceSwitcherStrategyManual)
+
+        self.assertEqual(switcher.services, self.services)
+        self.assertIsInstance(switcher.strategy, ServiceSwitcherStrategyManual)
+        self.assertEqual(switcher.strategy.services, self.services)
+
+    async def test_default_active_service(self):
+        """Test that the initially-active service receives frames while others don't."""
+        switcher = ServiceSwitcher(self.services, ServiceSwitcherStrategyManual)
+
+        # Reset counters
+        for service in self.services:
+            service.reset_counters()
+
+        # Send some test frames
+        frames_to_send = [
+            TextFrame(text="Hello 1"),
+            TextFrame(text="Hello 2"),
+            TextFrame(text="Hello 3"),
+        ]
+
+        await run_test(
+            switcher,
+            frames_to_send=frames_to_send,
+            expected_down_frames=[TextFrame, TextFrame, TextFrame],
+            expected_up_frames=[],  # Expect no error frames
+        )
+
+        # Only service1 should have processed the text frames
+        # Note: The service also receives StartFrame and EndFrame, so count those too
+        text_frames = [f for f in self.service1.processed_frames if isinstance(f, TextFrame)]
+        self.assertEqual(len(text_frames), 3)
+
+        # Check that other services don't receive text frames (they might get StartFrame/EndFrame)
+        service2_text_frames = [
+            f for f in self.service2.processed_frames if isinstance(f, TextFrame)
+        ]
+        service3_text_frames = [
+            f for f in self.service3.processed_frames if isinstance(f, TextFrame)
+        ]
+        self.assertEqual(len(service2_text_frames), 0)
+        self.assertEqual(len(service3_text_frames), 0)
+
+        # Verify the actual text frames processed
+        for i, frame in enumerate(text_frames):
+            self.assertEqual(frame.text, f"Hello {i + 1}")
+
+    async def test_service_switching(self):
+        """Test that after service switching using ManuallySwitchServiceFrame, the new active service receives frames while others don't."""
+        switcher = ServiceSwitcher(self.services, ServiceSwitcherStrategyManual)
+
+        # Reset counters
+        for service in self.services:
+            service.reset_counters()
+
+        # Send a test frame, a switch frame, and another test frame
+        await run_test(
+            switcher,
+            frames_to_send=[
+                TextFrame("Hello 1"),
+                ManuallySwitchServiceFrame(service=self.service2),
+                TextFrame("Hello 2"),
+            ],
+            expected_down_frames=[TextFrame, ManuallySwitchServiceFrame, TextFrame],
+            expected_up_frames=[],  # Expect no error frames
+        )
+
+        # Verify service2 received the frame
+        service1_text_frames = [
+            f for f in self.service1.processed_frames if isinstance(f, TextFrame)
+        ]
+        service2_text_frames = [
+            f for f in self.service2.processed_frames if isinstance(f, TextFrame)
+        ]
+        service3_text_frames = [
+            f for f in self.service3.processed_frames if isinstance(f, TextFrame)
+        ]
+
+        self.assertEqual(len(service1_text_frames), 1)
+        self.assertEqual(len(service2_text_frames), 1)
+        self.assertEqual(len(service3_text_frames), 0)
+
+        self.assertEqual(service1_text_frames[0].text, "Hello 1")
+        self.assertEqual(service2_text_frames[0].text, "Hello 2")
+
+    async def test_multi_service_switcher_targeting(self):
+        """Test that ManuallySwitchServiceFrame targets the correct ServiceSwitcher in a multi-switcher pipeline."""
+        # Create services for first switcher
+        switcher1_service1 = MockFrameProcessor("switcher1_service1")
+        switcher1_service2 = MockFrameProcessor("switcher1_service2")
+        switcher1_services = [switcher1_service1, switcher1_service2]
+
+        # Create services for second switcher
+        switcher2_service1 = MockFrameProcessor("switcher2_service1")
+        switcher2_service2 = MockFrameProcessor("switcher2_service2")
+        switcher2_services = [switcher2_service1, switcher2_service2]
+
+        # Create two service switchers
+        switcher1 = ServiceSwitcher(switcher1_services, ServiceSwitcherStrategyManual)
+        switcher2 = ServiceSwitcher(switcher2_services, ServiceSwitcherStrategyManual)
+
+        # Create a pipeline with both switchers: switcher1 -> switcher2
+        pipeline = Pipeline([switcher1, switcher2])
+
+        # Reset counters
+        for service in switcher1_services + switcher2_services:
+            service.reset_counters()
+
+        # Initially, both switchers should use their first services
+        self.assertEqual(switcher1.strategy.active_service, switcher1_service1)
+        self.assertEqual(switcher2.strategy.active_service, switcher2_service1)
+
+        # Send frames to test the pipeline:
+        # 1. Text frame (should go through both switchers' active services)
+        # 2. Switch frame targeting switcher1's second service
+        # 3. Text frame (should go through switcher1's new service and switcher2's original service)
+        # 4. Switch frame targeting switcher2's second service
+        # 5. Text frame (should go through switcher1's current service and switcher2's new service)
+        await run_test(
+            pipeline,
+            frames_to_send=[
+                TextFrame("Before any switches"),
+                ManuallySwitchServiceFrame(service=switcher1_service2),  # Switch first switcher
+                TextFrame("After switching first switcher"),
+                ManuallySwitchServiceFrame(service=switcher2_service2),  # Switch second switcher
+                TextFrame("After switching second switcher"),
+            ],
+            expected_down_frames=[
+                TextFrame,
+                ManuallySwitchServiceFrame,
+                TextFrame,
+                ManuallySwitchServiceFrame,
+                TextFrame,
+            ],
+            expected_up_frames=[],  # Expect no error frames
+        )
+
+        # Verify the active services changed correctly
+        self.assertEqual(switcher1.strategy.active_service, switcher1_service2)
+        self.assertEqual(switcher2.strategy.active_service, switcher2_service2)
+
+        # Verify frame distribution:
+        # First text frame should go through switcher1_service1 and switcher2_service1
+        switcher1_service1_texts = [
+            f for f in switcher1_service1.processed_frames if isinstance(f, TextFrame)
+        ]
+        switcher2_service1_texts = [
+            f for f in switcher2_service1.processed_frames if isinstance(f, TextFrame)
+        ]
+
+        # Second text frame should go through switcher1_service2 and switcher2_service1
+        switcher1_service2_texts = [
+            f for f in switcher1_service2.processed_frames if isinstance(f, TextFrame)
+        ]
+
+        # Third text frame should go through switcher1_service2 and switcher2_service2
+        switcher2_service2_texts = [
+            f for f in switcher2_service2.processed_frames if isinstance(f, TextFrame)
+        ]
+
+        # Verify frame counts and content
+        self.assertEqual(len(switcher1_service1_texts), 1)
+        self.assertEqual(switcher1_service1_texts[0].text, "Before any switches")
+
+        self.assertEqual(len(switcher1_service2_texts), 2)
+        self.assertEqual(switcher1_service2_texts[0].text, "After switching first switcher")
+        self.assertEqual(switcher1_service2_texts[1].text, "After switching second switcher")
+
+        self.assertEqual(len(switcher2_service1_texts), 2)
+        self.assertEqual(switcher2_service1_texts[0].text, "Before any switches")
+        self.assertEqual(switcher2_service1_texts[1].text, "After switching first switcher")
+
+        self.assertEqual(len(switcher2_service2_texts), 1)
+        self.assertEqual(switcher2_service2_texts[0].text, "After switching second switcher")
+
+
+if __name__ == "__main__":
+    unittest.main()

tests/test_utils_string.py

@@ -31,6 +31,7 @@ class TestUtilsString(unittest.IsolatedAsyncioTestCase):
         assert match_endofsentence("Valid scientific notation 1.23e4.") == 33
         assert match_endofsentence("Valid scientific notation 0.e4.") == 31
         assert match_endofsentence("It still early, it's 3:00 a.m.") == 30
+        assert match_endofsentence("That'll be $20.00.") == 18
         assert not match_endofsentence("This is not a sentence")
         assert not match_endofsentence("This is not a sentence,")
         assert not match_endofsentence("This is not a sentence, ")