Endpointing is insane! any idea how to fix this.

What is your smart endpointing set to?

To fix the high endpointing latency, try adjusting your

startSpeakingPlan

configuration:

json
"start_speaking_plan": {
  "waitSeconds": 0.8,
  "smartEndpointingEnabled": true
}

This configuration helps reduce perceived latency by optimizing how Vapi determines when a user has completed their turn. Source: - [Discord Channel](https://discord.com/channels/1211482211119796234/1331035264751112202)

High endpointing latency like 2231ms is usually caused by default silence timeouts (up to 1.5s) and added wait buffers, you can reduce it by lowering waitSeconds, enabling smart endpointing, and tweaking silence thresholds for faster turn-taking.