I imagine he likes Davendra Banhart, right?
I think the double-tracking of the vocal is the thing that lets this down though.
The way it's panned just sounds wrong.
I would try either:
1. Ditching the vocal double altogether and just picking one vocal take, & pan it firmly in the centre.
2. Keep the vocal double but turn it substantially down in volume - it's not on-point enough by a long way and the discrepancies are really distracting. And again, pan the main vocal right in the centre, and pan the radically quieter vocal double somewhere nearby-ish.
3. Same as above but only bring in the vocal double in the chorus ("Oh, cicada")
I'd also consider maybe compressing the vocal (harder? really hard?). Compression might be the wrong thing to do, but the vocal does seem slightly in-and-out, if you know what I mean - it's not quite in your face enough. If compression isn't the answer, then maybe brightening the vocal slightly might be, either by adding some air at 9 or 10K, or by subtracting a tiny bit of low-mid.
But get that vocal more central, and it may do the trick. Come to think of it, a further alternative to the double-tracked vocal - if you really think his voice requires thickening - might be a short delay. What vocal mic did you use, as a matter of interest?
Guitar sounds good though and the mix sounds well balanced. I might be tempted to add a very light foot-tap, bring it in on the 2nd verse or somthing and keep it there, to give the song a bit of propulsion. The "Oh cicada..." melody line is also crying out to me for a harmony part a major third up. Either on vocals or on a violin or harmonica.