Here we go...
http://img245.imageshack.us/img245/2302/fullsong.pngThis is the full song.
The program I used to open it is FL Studio. FL Studio has a "detect tempo" algorithm which told me that the song is 124 bpm. But let's not trust that one, let's check it ourselves.
First, we need a
reference point in the beat that we can work with. The bass drum isn't good for this (since low frequency = slow wave) so I've used the "snare drum" (I don't know if it's really a snare drum, but you hopefully know which one I mean).
A weird thing I have noticed about this song is that the snare drum shifts. If you divide the song in groups of 4 measures each (starting from the first measure to have drums), the drum starts to plays a tiny bit later in each measure. Starting with the next group however, it is spot-on again. This means we can only use the first snare in every 4 measures as a reference point.
So our first reference point is the first snare sound in the song, about 18 seconds into it.
http://img178.imageshack.us/img178/6930/firsth.pngNote that I've synced the grid in the background to 124bpm and matched the snare to one of the vertical lines (which are 1/16th of a beat apart).
Scrolling through the song, it shows that
all of our other reference points
(that is, every first snare in every 4 measures, as explained above) match the grid as well, which means: 124bpm must be correct. See for yourself:
http://img512.imageshack.us/img512/690/lastbeforequitepart.pngThis is the last reference point (57 seconds in) before what I call "the quiet part" starts (1:04 minutes in).
But the map was set to 124bpm in the first place. If that's the correct tempo, then why did it sound off?
Well, this is why:
http://img823.imageshack.us/img823/8994/firstafterquietpart.pngThis is the first reference point - the first snare -
after aforementioned quite part. You'll notice it's off.
But not only this one. Starting from this one, all of them are. So the bold "
all" I wrote earlier is only true until the quiet part.
However, I will now realign the song, so that this "first reference point after the quite part" matches the beat again:
http://img217.imageshack.us/img217/5272/firstafterquietpart3tic.pngSame position, same tempo (124bpm still), I just moved the song a bit to the left. Now it fits again. And so does the rest of the song. I'll prove it with the last reference point in the song (1:42 in):
http://img810.imageshack.us/img810/4760/last.pngThe conclusion of all of this is:
The song is indeed 124bpm, there is only a small delay/offset change somewhere in the quiet part. Because the last syllable Miku sings at 1:16 still fits the grid before I moved the song ...
http://img153.imageshack.us/img153/5141/lastmikusyllablebeforeq.png... we can narrow down the interval to 2 seconds. Yes I know, it doesn't fit
exactly, but she's the singer after all, not the drums, so that's to be expected. And it's a lot less than the first reference point after the quiet part was and in the other direction too.
Anyway, in those remaining 2 seconds, we have a piano chord. At 1:18.
http://img190.imageshack.us/img190/2658/thepianochord.pngAnd as you can see, this one is off quite a bit.
So..........
just like I said in
this post, the offset change is before the piano chord.
And 124bpm is correct.
For the sake of completeness, let's try 123.97bpm, like Alace said:
http://img37.imageshack.us/img37/7037/12397first.pngFirst reference point in the song: check.
http://img813.imageshack.us/img813/91/12397firstafterquietpar.pngFirst reference point after the quite part: check.
But that's it. All the others don't fit. The last reference point before the quite part, or the last one in the song, or any other one...
http://img651.imageshack.us/img651/4604/12397lastbeforequietpar.pnghttp://img299.imageshack.us/img299/738/12397last.pngSo the 123.97bpm kinda do compensate for the offset change, but aren't correct. 124 is the way to go.
There only remaining question is: how big exactly is this offset change?
And that's the single detail that I mentioned in my previous post, the one I might not have gotten right. I said "about 20ms", but let's calculate the correct value:
http://img217.imageshack.us/img217/2789/markers.pngI opened the song in FL Studio's wave editor and marked 3 reference points (please read the orange captions).
I marked them VERY precisely.
http://img6.imageshack.us/img6/2647/sampleperfect.pngAnd now, reading the positions from that editor:
"First" is @ 18.467 seconds
"Last before quiet part" is @ 57.176 seconds
"First after quiet part" is @ 80.418 seconds
"First" and "Last before quiet part" are 20 measures from each other in the song.
"First" and "First after quiet part" are 32 measures from each other in the song.
57.176 seconds - 18.467 seconds = 38.709 seconds
80.418 seconds - 18.467 seconds = 61.951 seconds
Now, extrapolating where "First after quiet part"
should be, if the tempo remained constant:
38.709 seconds / 20 measures * 32 measures = (38.709 / 20 * 32) seconds = 61.9344 seconds after "First"
Subtracting this from the actual distance between "First after quiet part" and "First", we get:
61.951 seconds - 61.9344 seconds = 0.0166 seconds = 16.6ms
And this is our offset change before the piano chord.
So please use 124bpm for your first timing section, and for the one that does not inherit the previous settings as well.
Set the offset for the latter to the offset of the first one plus 60.017.
edit: oops, little mistake corrected
. Sit tos SB arrowis ei kyl pahemmin mun mielest oo vikaa. Jos haluat korostaa sitä enemmän nii värii vois tummentaa.
