Speech designer works best with an audio track containing the words, and a text file containing the actual words.

IMHO. the best results for a singing voice (and accompanying music) would be to record yourself speaking the word in time with the music. For a singing song, you may expect to say some words fast; and some slowly. Use that recording with Speech Designer. After everything is done, simply replace the new recording with the original audio file.

In addition, it is worth looking at the phonemes files, as they show which phones are associated. Then, when creating the text file for Speech Designer, it is possible to mis-spell words deliberately in order to allow for effect.

I would be interested what the OP finds.