I’ll try to explain it using a translation. Simply put, the mouth opening/closing and mouth deformation parameters from facial tracking are mapped to the corresponding parameters in Live2D. The -1~0~1 values at the bottom represent the mouth expression, ranging from sad to smiling, while the 0~1 values on the side represent the mouth opening and closing.
Then, by putting it into VTS and creating vowel mouth shapes, you can see the exact positions of each mouth shape within the parameters.
The two mouths on the right: the ones above are the mouths modified using this method, while the ones below are the original versions before modification.