We're using LiveKit Python agents SDK v1.2.2 with AgentSession and Deepgram STT. We've set interim_results=False in our Deepgram STT configuration, but we're still seeing multiple transcript fragments being forwarded to room participants via text streams on the "lk.transcription" topic. For example, when a user says "It's pretty okay, I suppose. Quite sure. Just trying to talk a little bit so you can get some info.", we see: 1. First transcript event: "It's" 2. Second transcript event: "pretty okay, I suppose." 3. Third transcript event: "Quite sure." 4. Fourth transcript event: "Just trying to talk a little bit so you can get some info." The agent correctly aggregates these into a complete utterance before sending to the LLM, but the TranscriptSynchronizer appears to be forwarding each intermediate "final" transcript from Deepgram to the room participants. Is there a way to configure the TranscriptSynchronizer to only forward truly final transcripts (after end-of-utterance) rather than every transcript event from Deepgram? We want room participants to receive only complete utterances, not progressive fragments. Our current STT config: stt=deepgram.STT( model="nova-3", interim_results=False, endpointing_ms=25, punctuate=True, smart_format=True

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