text_channel_output = text_channel_output.split("<|eot|>")[0].replace(".....", "")

why speech to speech task should use replace(".....", "") ?

<img width="2268" height="454" alt="Image" src="https://github.com/user-attachments/assets/4f4cb368-c7b8-4160-8a4c-18a1bb536b61" />