When using LiveKit Cloud with OpenAI’s non-realtime STT models, you will not receive word-by-word or sentence-by-sentence (incremental) transcription. Instead, the transcribed text is delivered all at once after the full utterance is processed. This is a limitation of non-realtime models—interim or streaming transcription (needed for real-time display) is only supported when using a realtime STT provider or OpenAI’s realtime API with

use_realtime=true

. If you need real-time transcription for both user and agent (word or sentence level), you will need to: • Switch to a realtime STT model (either via OpenAI with real-time support, or another provider offering streaming transcription) • Configure your agents framework or client code to handle interim transcription events, which LiveKit emits on an interval when real-time STT is enabled Your current setup (STT-LLM-TTS with non-realtime models) cannot achieve this without a change to the transcription backend. There are no configuration flags available to make non-realtime models provide incremental transcription results. Let me know if you’d like guidance on enabling real-time STT for your flow or have questions about compatible providers. Sources: AgentTranscriptionOptions | LiveKit Agents | livekit-plugins/livekit-plugins-openai/livekit/plugins/openai/realtime/real... | Realtime model integrations | LiveKit Docs | AI voice agents | LiveKit Docs