Questions for Greetings

[gmltmd789]

Hello,

First, thank you for providing such excellent code.

I have three questions:

1. What is the exact role of the greetings? Is it intended to function as a prompt for zero-shot speaker adaptation?

2. Looking at the provided greeting file in jsonl format, it seems to be entirely in Chinese, and the video demo is also in Chinese. Does the model not support English conversations?

3. When decoding the jsonl file, there are many special tokens like <|empty|>. Could you clarify how the text and audio should be preprocessed to properly utilize the greetings?