[NM] 00:12:846 (12846|0) - 00:16:846 (16846|2) - These should be extended 1/3 past where they end for consistency with the LNs preceding them, and also cause the sound doesn't stop shorter than before.
00:18:179 (18179|1) - Should end 1/3 before where it currently does since there's a quick cut off between the synth and snare.
00:20:846 (20846|2,21513|2,22179|3,22846|3,23513|2,24179|2,24846|3,25513|3) - Since the kick seems to have a heavier weight, I'd suggest removing the whistles and letting the soft-hitnormal play instead to better match the weight of the percussion. 00:28:846 (28846|3,29179|3,29513|3,29846|3,30179|3) - ^
00:27:513 - I think you could vary the jumps here a bit since a 1/1 jumptrill is kind of boring and far too easy for a normal diff imo. Might try to stack one of the notes from each jump on each other like this https://osu.ppy.sh/ss/9910610
00:31:513 - There seems to be a disproportionate amount of LNs in every column but col 2. There's only 5 ish LNs in col 2 as opposed to 10-12 ish in all other columns. I'd try to rearrange some of the notes to avoid this uneven hand balance.
01:15:846 (75846|3,76068|0) - Might consider putting these in the same column cause they're the same pitch.
01:31:402 (91402|2) - Should be extended an extra third of a beat to 01:31:846 - as the sound is continuous. You've also connected every single LN to the note following it in this section so it's inconsistent to suddenly leave a disconnect.
[HD] 00:28:846 (28846|2,28846|3,29179|2,29179|3) - 00:29:846 (29846|0,29846|1,30179|0,30179|1) - More of a nitpick, but you haven't repeated the same 2 jumps here yet so I would change the placements of these jumps to avoid having a repeating, like this https://osu.ppy.sh/ss/9910672
00:51:513 (51513|2,51846|3,52179|3,52513|2) - Add drum-hitfinishes here to match the NM diff.
01:21:846 (81846|0) - I mean, it's fine to leave this as a 2/3 LN, but the sound is only 1/3 long with a short cut off on the next note at 01:21:957 (81957|3) - . Since it seems like you're going for musical accuracy on the LN lengths here, I would arrange this using capped LNs like so https://osu.ppy.sh/ss/9910701
01:27:179 (87179|0) - Similar to the above, it should be 1/3 long only.
01:28:179 (88179|2) - Make into a 2/3 long LN? Same goes for 01:30:846 (90846|0) - to act as a connect between both jumps.
I agree with ayyri about the bg, we want to keep the similarity with the original set 00:28:846 - imo current pattern is much more easier to read 01:21:846 - technically yes, they are 1/3 long but it's not really notice-able in normal speed (100% playback rate). it sounds more like 2/3 to me so I made them like that to simplify the pattern
Of all things, why the hell is 西む-42ab in tags... do you realise how general this is and how far from this song that tag is? That is a location. Furthermore you have 3 different sets of circles taking those 2 spots (booth location) every single biannual event