Is it possible to define speech duration as "Duration discriminator/Duration predictor" are a thing?

[Subarasheese]

Hello,

Can someone explain in layman terms what these are? Is it possible to force the speech to have an specific duration, like, X seconds? (the generated audio having that length and adapting the speech to fit it )

Thank you