Pipeline Nodes | SignalWire

These methods exist for API compatibility with LiveKit’s pipeline architecture. On SignalWire they are no-ops because the control plane handles the full STT/LLM/TTS pipeline automatically.

All three pipeline node methods are no-ops on SignalWire. They are provided solely so that code written for LiveKit’s pipeline architecture does not raise errors when run on SignalWire via LiveWire.

stt_node

1 from signalwire.livewire import Agent
2 
3 agent = Agent(instructions="You are a helpful assistant.", llm="gpt-4o")
4 await agent.stt_node(audio=None, model_settings=None)

No-op. SignalWire handles speech recognition in its control plane.

audio

AnyDefaults to None

Audio input. Accepted for API compatibility.

model_settings

AnyDefaults to None

STT model settings. Accepted for API compatibility.

Returns: None

llm_node

1 from signalwire.livewire import Agent
2 
3 agent = Agent(instructions="You are a helpful assistant.", llm="gpt-4o")
4 await agent.llm_node(chat_ctx=None, tools=None, model_settings=None)

No-op. SignalWire handles LLM inference in its control plane.

chat_ctx

AnyDefaults to None

Chat context. Accepted for API compatibility.

tools

AnyDefaults to None

Tool list. Accepted for API compatibility.

model_settings

AnyDefaults to None

LLM model settings. Accepted for API compatibility.

Returns: None

tts_node

1 from signalwire.livewire import Agent
2 
3 agent = Agent(instructions="You are a helpful assistant.", llm="gpt-4o")
4 await agent.tts_node(text=None, model_settings=None)

No-op. SignalWire handles text-to-speech in its control plane.

text

AnyDefaults to None

Text to synthesize. Accepted for API compatibility.

model_settings

AnyDefaults to None

TTS model settings. Accepted for API compatibility.

Returns: None