how can i log total latency at every time in my entrypoint LiveKit Community #ask-ai

Join Slack

how can i log total latency at every time in my en...

# ask-ai

cool-lizard-5130

05/15/2025, 12:58 PM

how can i log total latency at every time in my entrypoint

tall-airline-16756

05/15/2025, 12:58 PM

Thanks for reaching out, parag. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

tall-airline-16756

05/15/2025, 12:58 PM

To log latency in a LiveKit application, you can use the

metrics

module and subscribe to the

metrics_collected

event. Here’s an example implementation:

Copy code

# Import the metrics module
from livekit.agents import metrics

# Subscribe to metrics collection events
@agent.on("metrics_collected")
def _on_metrics_collected(mtrcs: metrics.AgentMetrics):
    # Use the helper to format and log based on metrics type
    metrics.log_metrics(mtrcs)

You can calculate the total conversation latency using the formula:

Copy code

total_latency = eou.end_of_utterance_delay + llm.ttft + tts.ttfb

Where: - end_of_utterance_delay: Time between VAD detecting end of speech and LLM inference. - ttft: Time to first token (LLM generating the first token). - ttfb: Time to first byte (TTS generating the first byte of audio). Let me know if you need further assistance! Sources: Capturing metrics | LiveKit Docs | Capturing metrics | LiveKit Docs

cool-lizard-5130

05/15/2025, 1:00 PM

will this work

Copy code

@session.on("metrics_collected")
    def _on_metrics_collected(ev: MetricsCollectedEvent):
        metrics = ev.metrics
        total_latency = (
            metrics.eou.end_of_utterance_delay + metrics.llm.ttft + metrics.tts.ttfb
        )
        <http://logger.info|logger.info>(f"current total latency - {total_latency}")
        metrics.log_metrics(metrics=ev.metrics, logger=logger)

tall-airline-16756

05/15/2025, 1:00 PM

Your implementation appears to align well with the general approach for handling the

metrics_collected

event in LiveKit. Here’s a breakdown:

Copy code

@session.on("metrics_collected")
def _on_metrics_collected(ev: MetricsCollectedEvent):
    metrics = ev.metrics
    total_latency = (
        metrics.eou.end_of_utterance_delay + metrics.llm.ttft + metrics.tts.ttfb
    )
    <http://logger.info|logger.info>(f"current total latency - {total_latency}")
    metrics.log_metrics(metrics=ev.metrics, logger=logger)

This should work as long as: - The

metrics

object contains the

eou

llm

, and

tts

attributes with the respective properties. - The

log_metrics

function is correctly implemented to handle the

metrics

object. If you encounter any issues, feel free to share the error or behavior, and I’ll assist further! Sources: Capturing metrics | LiveKit Docs | Capturing metrics | LiveKit Docs

2 Views

Open in Slack

Previous Next