Text this: VVC coded distortion prediction model based on frame-level transform coefficient modeling of generalized Gaussian distribution