Minor Addendum: Time to invoke some more loophole abuse, just for fun! The question does not explicitly state whose move it is. Also, traditionally, in helpmates, Black starts first. Thus, I think, the score is best expressed as the total number of White moves; which, at the moment, is a total of 31.
I invoke this mainly because the crux (read: whole) of my answer relies on it.
Seeing as how there aint't no rule against computers, and given that I am a chess composer who is aware of chess software, I went to my go-to helpmate solver, the Helpmate Analyzer--I'm glad to see that it has been found by others in the comments above/below.
After running the position through, we can down the score to 31.
The engine finds a unique helpmate in five, which is the only possible solution in fives moves, that involves knight promotion:
1.g5 g4 2.Qh5 gxh5 3.Nf7 h6 4.Rd8 h7 5.Bd7 hxg8=N#
This has the additional effect of verifying that 6 moves is indeed optimal for queen and rook promotion. The jury is still out on bishop promotion; although 7 moves seems optimal since there is no evidence, as of yet, against it.
For non-promotion, however, I can conjure up a weak proof of optimality. The king must walk over to support the pawn to give mate. Given the distance, the king is forced to walk 4 or 5 squares, in which the pawn must complement 1 or 2 moves. Intuition tells me that the pawn must move to f3, f4, f5, g4, or g3 for a potentially shorter mate, given the above math. However, Black can neither block off the king's escape squares and/or unprotect the mating squares in time. Thus, 7 moves is optimal for non-promotion.
And there we have it, a score of 31 with some proof attached.