As pointed out to me by HitCoder, it's possible to put plain arpabet phonemes into the notes of a UST, and have Arpasing Assistant convert them into diphones! If you leave them as-is, then diphones with a vowel as the second phoneme will be long notes, and everything else is short notes. If you add "/u" to the end, then in the same style as word input, it will divide the note evenly for each diphone.