Super Text Mesh just renders the data from the font directly, so not all fonts will be aligned with one another.
What I recommend is...
For text that uses a single font, you can edit the "baseOffset" variable to align the font how you like. But for multiple fonts, you can use the <y=offset> tag to offset text vertically.
So a string could look like... "ABCDEFG<f=chinese><y=0.2>你好</y></f>" (With 0.2 being whatever offset the text needs)
You can simplify this further by using Super Text Mesh's "Voices" data! If you create a voice named "chinese" with this text: "<f=chinese><y=0.2>", then all you have to type in your script is "ABCDEFG<v=chinese>你好</v>" !
