LLMUserAggregator broadcasts UserMuteStartedFrames before pushing StartFrames

[jjaju]

pipecat version

0.0.102

Python version

3.12.11

Operating System

WSL Ubuntu 24.04.3 LTS | python:3.12-bookworm, Debian 12 docker container

Issue description

When using the MuteUntilFirstBotCompleteUserMuteStrategy the LLMUserAggregator broadcasts a UserMuteStartedFrame before pushing the StartFrame to downstream processors (e.g. LLMService), causing an ERROR Log:

| ERROR    | pipecat.processors.frame_processor:_check_started:946 - YourLLMService#2 Trying to process UserMuteStartedFrame#6 but StartFrame not received yet

Reproduction steps

Configure a pipeline with MuteUntilFirstBotCompleteUserMuteStrategy:

context = LLMContext()
context_aggregator = LLMContextAggregatorPair(
    context,
    assistant_params=LLMAssistantAggregatorParams(),
    user_params=LLMUserAggregatorParams(
        user_mute_strategies=[
            MuteUntilFirstBotCompleteUserMuteStrategy(),
        ],
        vad_analyzer=SileroVADAnalyzer(
            params=VADParams(
                stop_secs=0.2,
            )
        ),
        user_idle_timeout=12.0,
    ),
)
pipeline = Pipeline(
    [
        transport.input(),
        stt,
        context_aggregator.user(),
        llm,
        tts,
        transport.output(),
        context_aggregator.assistant(),
    ]
)

Start the pipeline and observe logs.

Expected behavior

StartFrame reaches all processors before UserMuteStartedFrame.

Actual behavior

UserMuteStartedFrame reaches downstream processors before StartFrame

Logs

| ERROR    | pipecat.processors.frame_processor:_check_started:946 - YourLLMService#2 Trying to process UserMuteStartedFrame#6 but StartFrame not received yet