https://livekit.io logo
Join Slack
Powered by
how can I check for how muh time the the on_user_t...
# ask-ai
a
how can I check for how muh time the the on_user_turn_completed didn't change
t
Thanks for reaching out, Aditya. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.
There isn’t a built-in LiveKit method to directly check how much time has passed since 
on_user_turn_completed
last changed. You’ll need to capture this timing yourself by storing a timestamp each time 
on_user_turn_completed
is called, and then calculate the difference using the current time:
Copy code
import time
from livekit.agents import Agent
from livekit.agents.llm.chat_context import ChatContext, ChatMessage

class MyAgent(Agent):
    def __init__(self):
        super().__init__()
        self.last_user_turn_completed_time = None
    
    async def on_user_turn_completed(self, turn_ctx: ChatContext, new_message: ChatMessage) -> None:
        self.last_user_turn_completed_time = time.time()
        # Your logic here
    
    def get_time_since_last_turn_completed(self) -> float | None:
        if self.last_user_turn_completed_time is None:
            return None
        return time.time() - self.last_user_turn_completed_time
Alternatively, you can use LiveKit’s End-of-Utterance (EOU) metrics for more detailed event timing. Listen for EOU metric events, save the 
timestamp
, and use that for your calculations. The 
EOUMetrics
object also provides fields like 
on_user_turn_completed_delay
and 
end_of_utterance_delay
for further insight. This approach requires maintaining state within your agent, as there’s no built-in query for this exact information. More detail on these methods and available EOU metrics can be found in the LiveKit documentation. Sources: livekit-plugins/livekit-plugins-google/livekit/plugins/google/beta/realtime... | Capturing metrics | LiveKit Docs
a
how can I use EOU there?
t
Hey there! Let me walk you through how you can track user turn timings in LiveKit using End-of-Utterance (EOU) metrics. It's super easy – just subscribe to the 
eou_metrics_collected
event on your speech-to-text component. Think of it like getting a behind-the-scenes peek at how conversations flow!
Copy code
class MyAgent(Agent):
    def __init__(self):
        super().__init__()

        def eou_wrapper(metrics):
            asyncio.create_task(self.on_eou_metrics_collected(metrics))
        self.stt.on("eou_metrics_collected", eou_wrapper)

    async def on_eou_metrics_collected(self, metrics):
        print(f"EOU timestamp: {metrics.timestamp}")
        print(f"End of utterance delay: {metrics.end_of_utterance_delay}")
        print(f"Transcription delay: {metrics.transcription_delay}")
        print(f"on_user_turn_completed_delay: {metrics.on_user_turn_completed_delay}")
In the 
on_eou_metrics_collected
method, you'll get access to some really cool timing data that can help you understand conversation dynamics. • `timestamp`: When the EOU event happened • `end_of_utterance_delay`: How long after someone stops speaking • 
transcription_delay
and `on_user_turn_completed_delay`: Handy latency details Just a heads up – this works with voice activity detection (VAD) or LiveKit's turn detector, but not with server-side turn detection. It's a fantastic way to get more insight into how conversations are unfolding. Check out the example above to see how you can customize your own handler and play around with these metrics! Sources: Capturing metrics | LiveKit Docs | livekit-agents/livekit/agents/metrics/base.py | complex-agents/turn-taking/agent.py | Capturing metrics | LiveKit Docs
Open in Slack
PreviousNext
Powered by