How to change the value of "Silence threshold" before training?

[upColor]

When doing inference through the GUI after training, I always have to manually set the "Silence threshold" = -60.0 (the lowest value), otherwise (with the default value) missing words or missing pronunciation can frequently occur within the generated audio.

But before training, how to the value of the "Silence threshold" please? Does it have a controlling parameter, like:
svc pre-hubert -st -60.0