Base Service Classes

Base class for transport implementations that handle audio/video I/O streams.

Name identifier for the input stream

Name identifier for the output stream

Event loop to use for async operations. Defaults to the current running loop.

Returns the input frame processor. Must be implemented by subclasses.

@abstractmethod def input(self) -> FrameProcessor

Returns the output frame processor. Must be implemented by subclasses.

@abstractmethod def output(self) -> FrameProcessor 

Decorator to register an event handler for the specified event.

def event_handler(self, event_name: str)

Adds an event handler function for the specified event.

def add_event_handler(self, event_name: str, handler)

Configuration parameters for transport audio/video capabilities.

Enable camera output

Whether camera output is live streaming

Camera output width in pixels

Camera output height in pixels

Camera output bitrate in bits/second

Camera output framerate

Camera output color format

Enable audio output

Whether audio output is live streaming

Audio output sample rate in Hz

Number of audio output channels

Audio output bitrate in bits/second

Audio mixer for output processing

Enable audio input

Audio input sample rate in Hz

Number of audio input channels

Audio filter for input processing

Enable Voice Activity Detection

When the VAD is enabled, determines whether to pass through audio frames. This is required in order to pass through audio if you’re using an STT service and have the VAD enabled.

Voice Activity Detection analyzer. You can set this to either SileroVADAnalyzer() or WebRTCVADAnalyzer().

Base class for Speech-to-Text services.

Whether to pass through audio frames

Sets the STT model.

@abstractmethod async def set_model(self, model: str)

Sets the recognition language.

@abstractmethod async def set_language(self, language: Language)

Base class for LLM services.

Registers a function handler. Use None as function_name to handle all functions.

def register_function( self, function_name: str | None, callback,
start_callback=None )

Removes a registered function handler.

def unregister_function(self, function_name: str | None)

Checks if a function handler is registered.

def has_function(self, function_name: str) -> bool

Requests an image from a user.

async def request_image_frame(
    self,
    user_id: str,
    *,
    text_content: str | None = None
)

Base class for Text-to-Speech services.

Output audio sample rate in Hz

Text filter for preprocessing

Sets the TTS model.

@abstractmethod async def set_model(self, model: str)

Sets the voice identifier.

@abstractmethod def set_voice(self, voice: str)

Converts text to speech immediately.

async def say(self, text: str)

Base class for text preprocessing filters used by TTS services.

Removes Markdown formatting from text while preserving structure and readability.

from pipecat.utils.text import MarkdownTextFilter

text_filter = MarkdownTextFilter(
    params=MarkdownTextFilter.InputParams(
        enable_text_filter=True,
        filter_code=True,
        filter_tables=True
    )
)

Enable/disable text filtering

Remove code blocks from the text

Remove Markdown tables from the text

Base class for Image Generation services.

Core image generation method to be implemented by services.

@abstractmethod
async def run_image_gen(self, prompt: str) -> AsyncGenerator[Frame, None]

Base class for Vision services.

Core vision processing method to be implemented by services.

@abstractmethod
async def run_vision(
    self,
    frame: VisionImageRawFrame
) -> AsyncGenerator[Frame, None]