LiveKit Community

total_latency = end_of_utterance_delay + ttft + ttfb

whats a good total latency value

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


Total latency in LiveKit is calculated as the sum of _end_of_utterance_delay_, _ttft_ (Time to First Token), and _ttfb_ (Time to First Byte). I understand how important low latency is for smooth, real-time interactions, though there aren't specific benchmarks for what counts as a "good" total latency value just yet.
If you'd like, I’m here to help you measure or optimize these latency components in your application—just let me know how I can support you!
 Sources: <https://github.com/livekit/agents/blob/f2fb5705c865ee7f220f60b5baeaf4c716db4cf6/livekit-agents/livekit/agents/metrics/utils.py#L1-L55|livekit-agents/livekit/agents/metrics/utils.py> | <https://deepwiki.com/livekit/livekit_composite/2.2-speech-and-audio-processing|Speech &amp; Audio Processing | livekit/livekit_composite | DeepWiki> | <https://deepwiki.com/livekit/livekit_composite/2.5-agent-testing-and-evaluation|Agent Testing &amp; Evaluation | livekit/livekit_composite | DeepWiki> | <https://docs.livekit.io/agents/v0/build/metrics/|Capturing metrics | LiveKit Docs> | <https://docs.livekit.io/agents/v0/build/metrics|Capturing metrics | LiveKit Docs>