service_attributes

Functions for adding attributes to OpenTelemetry spans.

This module provides specialized functions for adding service-specific attributes to OpenTelemetry spans, following standard semantic conventions where applicable and Pipecat-specific conventions for additional context.

pipecat.utils.tracing.service_attributes.add_tts_span_attributes(span: Span, service_name: str, model: str, voice_id: str, text: str | None = None, settings: ServiceSettings | None = None, character_count: int | None = None, operation_name: str = 'tts', ttfb: float | None = None, **kwargs) → None[source]

Add TTS-specific attributes to a span.

Parameters:

  • span – The span to add attributes to.

  • service_name – Name of the TTS service (e.g., “cartesia”).

  • model – Model name/identifier.

  • voice_id – Voice identifier.

  • text – The text being synthesized.

  • settings – Service configuration settings.

  • character_count – Number of characters in the text.

  • operation_name – Name of the operation (default: “tts”).

  • ttfb – Time to first byte in seconds.

  • **kwargs – Additional attributes to add.

pipecat.utils.tracing.service_attributes.add_stt_span_attributes(span: Span, service_name: str, model: str, operation_name: str = 'stt', transcript: str | None = None, is_final: bool | None = None, language: str | None = None, user_id: str | None = None, settings: ServiceSettings | None = None, vad_enabled: bool = False, ttfb: float | None = None, **kwargs) → None[source]

Add STT-specific attributes to a span.

Parameters:

  • span – The span to add attributes to.

  • service_name – Name of the STT service (e.g., “deepgram”).

  • model – Model name/identifier.

  • operation_name – Name of the operation (default: “stt”).

  • transcript – The transcribed text.

  • is_final – Whether this is a final transcript.

  • language – Detected or configured language.

  • user_id – User ID associated with the audio being transcribed.

  • settings – Service configuration settings.

  • vad_enabled – Whether voice activity detection is enabled.

  • ttfb – Time to first byte in seconds.

  • **kwargs – Additional attributes to add.

pipecat.utils.tracing.service_attributes.add_llm_span_attributes(span: Span, service_name: str, model: str, stream: bool = True, messages: str | None = None, output: str | None = None, tools: str | None = None, tool_count: int | None = None, tool_choice: str | None = None, system_instructions: str | None = None, parameters: dict[str, Any] | None = None, extra_parameters: dict[str, Any] | None = None, ttfb: float | None = None, **kwargs) → None[source]

Add LLM-specific attributes to a span.

Parameters:

  • span – The span to add attributes to.

  • service_name – Name of the LLM service (e.g., “openai”).

  • model – Model name/identifier.

  • stream – Whether streaming is enabled.

  • messages – JSON-serialized messages.

  • output – Aggregated output text from the LLM.

  • tools – JSON-serialized tools configuration.

  • tool_count – Number of tools available.

  • tool_choice – Tool selection configuration.

  • system_instructions – System instructions.

  • parameters – Service parameters.

  • extra_parameters – Additional parameters.

  • ttfb – Time to first byte in seconds.

  • **kwargs – Additional attributes to add.

pipecat.utils.tracing.service_attributes.add_gemini_live_span_attributes(span: Span, service_name: str, model: str, operation_name: str, voice_id: str | None = None, language: str | None = None, modalities: str | None = None, settings: ServiceSettings | None = None, tools: list[dict] | None = None, tools_serialized: str | None = None, transcript: str | None = None, is_input: bool | None = None, text_output: str | None = None, audio_data_size: int | None = None, **kwargs) → None[source]

Add Gemini Live specific attributes to a span.

Parameters:

  • span – The span to add attributes to.

  • service_name – Name of the service.

  • model – Model name/identifier.

  • operation_name – Name of the operation (setup, model_turn, tool_call, etc.).

  • voice_id – Voice identifier used for output.

  • language – Language code for the session.

  • modalities – Supported modalities (e.g., “AUDIO”, “TEXT”).

  • settings – Service configuration settings.

  • tools – Available tools/functions list.

  • tools_serialized – JSON-serialized tools for detailed inspection.

  • transcript – Transcription text.

  • is_input – Whether transcript is input (True) or output (False).

  • text_output – Text output from model.

  • audio_data_size – Size of audio data in bytes.

  • **kwargs – Additional attributes to add.

pipecat.utils.tracing.service_attributes.add_openai_realtime_span_attributes(span: Span, service_name: str, model: str, operation_name: str, session_properties: dict[str, Any] | None = None, transcript: str | None = None, is_input: bool | None = None, context_messages: str | None = None, function_calls: list[dict] | None = None, tools: list[dict] | None = None, tools_serialized: str | None = None, audio_data_size: int | None = None, **kwargs) → None[source]

Add OpenAI Realtime specific attributes to a span.

Parameters:

  • span – The span to add attributes to.

  • service_name – Name of the service.

  • model – Model name/identifier.

  • operation_name – Name of the operation (setup, transcription, response, etc.).

  • session_properties – Session configuration properties.

  • transcript – Transcription text.

  • is_input – Whether transcript is input (True) or output (False).

  • context_messages – JSON-serialized context messages.

  • function_calls – Function calls being made.

  • tools – Available tools/functions list.

  • tools_serialized – JSON-serialized tools for detailed inspection.

  • audio_data_size – Size of audio data in bytes.

  • **kwargs – Additional attributes to add.