This is great! You control the pacing incredibly well! The blend of sounds is unexpected, but it works.

I'd consider messing around with the vocals a bit. They sound a little synthetic (especially in the faster sections). This is mostly due to the way that the sample fades in and out; you can hear individual attacks and releases of notes. I'd consider adding a little reverb to the vocals to mask this effect a bit. It should smooth out the gaps between individual notes as well as give the end of notes a more natural sounding taper. There are limits to this effect, though. Obviously if you crank the reverb too much, it'll make it sound like the choir is in a cement room a mile away. You can also try nudging some of the notes to start a bit earlier. This way you don't hear the end of one note, then the beginning of another, but they'll sort of crossfade and blend together a bit more. If you're able to do this with your particular sample set, you can also try lengthening the release time of your sample. Not all instruments give you access to the ADSR, but it's something I'd look for. I mess with this an awful lot in my own tracks. The slow attack on slow sections sounds great, but sometimes you really need notes to pop, so I'll automate the attack time to be quicker for those sections (or duplicate the track with different settings and chop the sequence between the two...it depends mostly on how big the project already is). Lastly, think about adjusting the velocities of each note in the vocal line. There are some notes that pop out a little unnaturally. Shaping the contour of a line is, to me, one of the most difficult things to do. You do a really good job of this already, but there are a few notes that still pop out oddly to my ears.

I hope I don't sound like a total twat. I think the piece is really well written and it works as is. These are just ideas to take or leave. Thank you for sharing this!