How to define stream chunk size?

[mx305]

Checks

• This template is only for usage issues encountered.
• I have thoroughly reviewed the project documentation but couldn't find information to solve my problem.
• I have searched for existing issues, including closed ones, and couldn't find a solution.
• I am using English to submit this issue to facilitate community communication.

Environment Details

ubuntu 22.04, Python 3.10, torch 2.4.0+cu124

Steps to Reproduce

What recommendations can be made for max_chars \ few_chars \ min_chars? Text chunking is performed via the chunk_text method.
Currently, in Chinese text testing, if the above three parameters are all fixed at 60, which results in occasional missing characters at the beginning of chunks.
If I directly calculate it using the default logic in socket_server.py, some garbled characters (a few random characters) appear at the end of chunks.

ref_text_byte_len = len(self.ref_text.encode("utf-8"))
self.max_chars = int(ref_text_byte_len / (ref_audio_duration) * (25 - ref_audio_duration))
self.few_chars = int(ref_text_byte_len / (ref_audio_duration) * (25 - ref_audio_duration) / 2)
self.min_chars = int(ref_text_byte_len / (ref_audio_duration) * (25 - ref_audio_duration) / 4)

Additionally: When few_chars and min_chars in socket_server.py are processed consecutively for first sentence, could this cause the earlier chunks to be split into overly small pieces?
Thanks you!

✔️ Expected Behavior

No response

❌ Actual Behavior

No response