I finally generated a new track today - so it has my own voice, but annoyingly not the original melodies. Although to be fair the AI got them fairly close. The original was written (by myself in the late sixties.
Something I have realised in my experiments with Suno, something I should have realised is that not all lyrics are equal when faced with different genres. The time signature, the beat, the groove, whatever may not "fit" the genre - or vice versa!. This is why the AI will sometimes leave out a word or double up on a phrase, etc. The lyrics must fit the genre/style.
I think this is why when I have problems selecting a genre, jazz is often a good bet to cope with the lyric. It has the capacity to "fit" any melodies to whatever lyrics are provided. I guess you could call it musical freedom. One other issue made this a difficult lyric for the AI is that the lines are not of equal length; close but not exact. When I sing the song as written, I usually take quite a free and easy position. Simply strumming along with my vocal allows me to be "free" to sing the words as written.
I am learning something every time (almost) that i use Suno to create a new track from an old song. I hope you like this one.