Special characters in the Text-To-Generate field which change outputs

[JohnH-128]

Sometimes I am not happy with the infering output. The slicing of the sentences are not always ok. Some texts I would like to put into the next step (or audio clip). Is that somehow possible to use some special character or combintaion in the Text-To-Generate prompt to put the text after the character in new audio? I started to use the "bla bla" text, but it takes a lot of test, and sometimes F5-TTS still slices the wrong place.
If there are modifier special characters or combinations, Where can I find the full list of them?
Last question: for an AI there is a "Temperature" setup for different result. How can I change the result of the F5-TTS ?

[JohnH-128]

[SWivid]

1. tab Multi-Style / Multi-Speaker Generation
2. use 2 styles but same ref_audio
3. interleave style1_text style2_text ..., for explicit split