Max sample sequence
bug
### Checks
- [x] This template is only for bug reports, usage problems go with 'Help Wanted'.
- [x] I have thoroughly reviewed the project documentation but couldn't find information to solve my problem.
- [x] I have searched for existing issues, including closed ones, and couldn't find a solution.
- [x] I am using English to submit this issue to facilitate community communication.
### Environment Details
pass
### Steps to Reproduce
pass
### ✔️ Expected Behavior
https://github.com/SWivid/F5-TTS/blob/c279a2b7d573bc87a4eac6f6c08882188dea1b10/src/f5_tts/model/backbones/dit.py#L46
https://github.com/SWivid/F5-TTS/blob/c279a2b7d573bc87a4eac6f6c08882188dea1b10/src/f5_tts/model/cfm.py#L95
Maybe we need to set CFM sample max_duration=8192. Cause when we train the model, it will sample at 5w steps. If use 4096 as the max_duration, when the prompt > 2048 frames, the _ref.wav length is not equal _gen.wav
https://github.com/SWivid/F5-TTS/blob/c279a2b7d573bc87a4eac6f6c08882188dea1b10/src/f5_tts/model/trainer.py#L413-L430
### ❌ Actual Behavior
👆
0 条评论