I've done a couple of calls using Vapi assistant, and it seems that of the connected calls, some of the users can't hear the assistant's output. in addition, the assistant seems to be misinterpreting the inputs, and unable to complete sentences that it was given. The latency is also super high End user unable to hear: Call ID: 5e43cbc7-89af-4a0e-a7cd-67b6d868b1d4 (timestamp: 07:07:19:777) it is supposed to finish its sentences but doesn't. The logs show that the model has output, but it is not translated to voice. Call ID: afd19cdd-49f4-47f0-84df-7bb78ef06953 (timestamp: 07:20:15:399). Latency is 4008ms, user cannot hear. Misinterpreting context: Call ID: 02b2be92-2a57-4448-9aa0-69cdbbbe20b8 (timestamp: 07:07:20:130), it is supposed to say "if you don't like what you're hearing, feel free to cut me off anytime"

hey @uncertainj thanks for sharing these call_ids let me review them and then get back to you.