I have an issue with Whisper+Grok LLM Request Delay = 0.1s 1. I said "Would you like to know my name?" 2. It recognised "Would you like to?" and "Would you like to know my name?" 3. It sends promt for each of this two 4. It starts play response for the first 5. Then second response start to interrupt speach of the first and the whole experience is always interrupted PS: I'm not able to set "LLM Request Delay" equal to 0.5 sec https://dashboard.vapi.ai/calls/37c73159-bfc5-4fe5-8f64-59beb5c3ab9c https://cdn.discordapp.com/attachments/1248309482472079493/1248309482753232931/image.png?ex=6663324d&is=6661e0cd&hm=0c116e2f58541ea6ffe43d9ccc5298b1e669822998f849385957309fa1232f87& https://cdn.discordapp.com/attachments/1248309482472079493/1248309483005022360/image.png?ex=6663324d&is=6661e0cd&hm=8581d137e523baa1d286bf0d9aa94cd0335022ba2c33780d0c6dfb6311ba40af&

@wehead_ilya I understand the issue you're facing with Whisper+Grok where the responses are overlapping due to the system recognizing two separate prompts from a single utterance. This can indeed create a confusing experience. To address this, we need to ensure that the system is configured to handle overlapping prompts more gracefully. Since you mentioned that you're unable to set the "LLM Request Delay" to 0.5 seconds, which could potentially resolve the overlapping responses, we should look into why that setting isn't available or not working as expected. Could you please provide more details on the steps you've taken to try and change the "LLM Request Delay" setting? Additionally, are there any specific error messages or issues you encounter when attempting to adjust this setting? This information will help me assist you more effectively.

Can you share some latest call_id?

I also can't set this to 0.5 as the number is not editable and the slider goes between whole numbers. Once moved from 0 to 1 or 2 it will never even go back to the default 0.1

Create a new thread and share your assistant_id and the things you need to change. I will do it for you.

hi @Sahil ive shared the ID 37c73159-bfc5-4fe5-8f64-59beb5c3ab9c of the conversation I've observed the problem

We process things word by word, then we wait for the confirmation of the final output from Whisper, and finally send it to TTS