To enable logging in your LiveKit agent boilerplate code and monitor components like STT, TTS, and LLM, follow these steps: 1. Import the Metrics Module:

from livekit.agents import metrics

2. Set Up Logging:

import logging
logger = logging.getLogger("agent-metrics")
logger.setLevel(<http://logging.INFO|logging.INFO>)

3. Subscribe to Metrics Events: Use the

metrics_collected

event to capture metrics for each component:

@agent.on("metrics_collected")
def _on_metrics_collected(ev: metrics.MetricsCollectedEvent):
    metrics.log_metrics(ev.metrics, logger=logger)

4. Log Specific Metrics: For detailed logging of each component:

@agent.on("metrics_collected")
def _on_metrics_collected(ev: metrics.MetricsCollectedEvent):
    if isinstance(ev.metrics, metrics.STTMetrics):
        <http://logger.info|logger.info>(f"STT Metrics: Audio Duration: {ev.metrics.audio_duration}, Streamed: {ev.metrics.streamed}, Error: {ev.metrics.error}")
    elif isinstance(ev.metrics, metrics.TTSMetrics):
        <http://logger.info|logger.info>(f"TTS Metrics: Audio Duration: {ev.metrics.audio_duration}, TTFB: {ev.metrics.ttfb}")
    elif isinstance(ev.metrics, metrics.LLMMetrics):
        <http://logger.info|logger.info>(f"LLM Metrics: TTFT: {ev.metrics.ttft}, Tokens per Second: {ev.metrics.tokens_per_second}, Total Tokens: {ev.metrics.total_tokens}")

5. Aggregate Metrics: Use the

UsageCollector

class to aggregate metrics across a session:

usage_collector = metrics.UsageCollector()
@agent.on("metrics_collected")
def _on_metrics_collected(ev: metrics.MetricsCollectedEvent):
    usage_collector.collect(ev.metrics)
async def log_usage():
    summary = usage_collector.get_summary()
    <http://logger.info|logger.info>(f"Usage Summary: {summary}")

For more advanced use cases, such as sending metrics to a server, you can modify the event handler to send data to an HTTP endpoint. Let me know if you need further assistance!