The voice agent sometimes says phrases twice, how can I solve this? https://cdn.discordapp.com/attachments/1291040814277070848/1291040814524530738/image.png?ex=66fea6f4&is=66fd5574&hm=96f5e6d89e5ea98256bd4929975df37accd3bca78de96d40ba47954b55439648& https://cdn.discordapp.com/attachments/1291040814277070848/1291040814763343977/image.png?ex=66fea6f4&is=66fd5574&hm=0c19be40dc75731fcf3bff4cf295486e6dee8bec9c04199b10f4211311b325ca&

@BrightMan I'm sorry to hear that you're experiencing issues with repeated phrases. This could be due to a technical glitch or a misconfiguration in the settings. To help resolve this, could you please provide me with more details? Specifically, it would be helpful to know if this happens with specific phrases or in certain situations, and if there have been any recent changes to the system or settings that might have affected the voice agent's behavior.

Here are entries from the log: 10:10:09:717 [LOG] Transcriber output: My name is Stewart, 10:10:09:732 [LOG] Endpointing timeout 2496ms 10:10:10:280 [CHECKPOINT] Assistant speech started 10:10:10:280 [INFO] Turn latency: 562ms (transcriber: 0ms, endpointing: 0ms, model: 0ms, voice: 0ms) 10:10:12:230 [CHECKPOINT] Model request started 10:10:12:233 [LOG] Model request started (attempt #1, gpt-4o-mini-2024-07-18, azure-openai, eastus) 10:10:12:439 [LOG] Model output: Thanks 10:10:12:439 [CHECKPOINT] Model sent first output token 10:10:12:439 [CHECKPOINT] Model sent start token 10:10:12:446 [LOG] Model output: , 10:10:12:450 [LOG] Model output: Stewart 10:10:12:457 [LOG] Model output: ! 10:10:12:465 [LOG] Model output: Could 10:10:12:468 [LOG] Model output: you 10:10:12:476 [LOG] Model output: please 10:10:12:481 [LOG] Model output: spell 10:10:12:489 [LOG] Model output: that 10:10:12:497 [LOG] Model output: for 10:10:12:499 [LOG] Model output: me 10:10:12:507 [LOG] Model output: ? 10:10:12:514 [LOG] Model request cost (attempt #1, $0.0003741, 2446 prompt, 12 completion) 10:10:12:514 [LOG] Voice input: Thanks, Stewart! Could you please spell that for me? 10:10:12:514 [CHECKPOINT] Model sent end token 10:10:16:407 [CHECKPOINT] Assistant speech stopped But the transcript shows:

@BrightMan can you share the call id?

a7a540a2-5022-47ae-b804-61aa2784bd9b

Did you get a solution already?

Still not solved

@BrightMan apollogy for the delay and somehow this thread go lost, i checked into both call ids. To me it's from prompting, I suggest following vapi prompting guide and create a diagram/board to track your assistant flow better.

Thank you, i'll try that.