Thanks — quick acknowledgement: you're asking whether adaptive interruption handling requires LiveKit Cloud Inference when your model is deployed to the cloud but you’re using a direct STT SDK plugin. Short answer: No — using LiveKit Cloud Inference is not strictly required. Adaptive interruption handling requires: • your agent is deployed to LiveKit Cloud (or running in dev mode), • Voice Activity Detection (VAD) is enabled, • the LLM is not a realtime model, and • the STT provides aligned transcripts. All LiveKit Inference STT models do provide aligned transcripts. If you use a direct STT SDK plugin, adaptive interruption handling will work only if that plugin exposes aligned transcripts (check

stt.capabilities.aligned_transcript

or

stt.capabilities.alignedTranscript

). If that capability is false or unavailable, adaptive interruption handling will not be supported. If you want, tell me which STT plugin you’re using and I’ll walk you through how to check its

capabilities

or suggest next steps. Sources: Adaptive interruption handling | LiveKit Documentation | Turns overview | LiveKit Documentation, with interactive elements