Forums » Songs » I have a timing problem with a song I'm making. Help would be nice

1
quite some time ago, I had a similar thread where I was having timing problems and once again I have a similar problem. it's an odd one. the song I am having trouble with is "Get The Gold" by Caramella Girls ( I'll post a download link to the file further down) and I am almost certain I have the offset correct and the main bpm correct too (140.1 bpm). my problem is there is a bpm change later in the song at the 79 measure marker (around 134 seconds in) and it's odd cause it's not a big change I currently have the changed bpm at 139.7 as opposed to 140.1 which is NOT much of a beat change but it sounds good but not like I want it. its not perfect and possibly with some more tweaking I could get it correct but my other bigger problem is I have no idea when the beat changes back to 140.1. it's weird and I'm not sure what to do really. there is no good program (or any that I know of) that can detect tempo changes in a song and I'm not sure what the changed bpm is supposed to be or if I even have the main bpm 100% correctly (also the offset may be off by a tiny bit but I would have figured that out once I actually started placing the notes. I literally just started making this song.) I am actually now just branching into songs with bpm changes. my last one had one in too but that one was obvious as hell. literally half the bpm xD

I haven't laid any notes besides some simple timing notes at the bpm change so I could switch on clap/metronome see if things lined up correctly or not.

http://www.mediafire.com/download/ndv2a5dhev5o1qy/Caramella+Girls-Get+The+Gold.rar

while were on the subject, how do you guys figure out tempo changes? just guess/figure it out on your own or...

also if/when I finish this song (and maybe redo my very old first one that sucks... badly) would people be interested in the few pad simfiles I've created as a pack? they are only expert... well... one difficulty and they have no artwork at all except for one which does and another who only has the banner (both just screen caps resized to fit correctly). they're just songs that have gotten stuck in my head so I made simfiles for them. they include every caramella girls song except caramelldansen english version (which is my really old one I may redo), Spooky Scary Skeletons (Remix) by The Living Tombstone, town of salem rap by Boyinaband, Diddley Dee by Cartoons, and I have some unfinished ones I will either finish (probably not), or not include. if so, should I attempt to make artwork (I am a honorable artist) or keep the artwork blank?

just do whatever you want with the simfile to help me if you want and you don't have to actually place correct notes, I just need help with timing/tempo. much appreciated :)
anyways thank you for the help
-Tal-

Last edited: 25 June 2015 1:12am

Reply
while were on the subject, how do you guys figure out tempo changes? just guess/figure it out on your own or...

I don't really have a scientific approach to finding BPM (I can do it by trial and error and by ear faster than I could ever do it mathematically) but I'll lay out my technique which does get me good results. It requires a good sense of rhythm and decent keyboard Stepmania skill (the ability to get a very low great count / Perfect(Marvelous?) Attack on the song you're working on).

I don't ever go into wave editing tools for the purpose of getting the BPM. I use FL Studio (which can play a metronome at a BPM to a precision of 0.001BPM, which is why it's awesome!) for getting the BPM initially:
I play the music file in some media player program and turn on the metronome in FL Studio. I tinker with the BPM. This step sounds more complicated than it is and it often takes less than a few minutes. I stop tinkering when one of 2 things happens:

1) The piece of music has a constant BPM and I've found exactly what it is by ear, listening to the metronome compared to the music playing. Note that verifying the BPM is constant requires turning on the metronome at the beginning of the song and letting it play along with the music until the end of the song. Most of the time you can tell when a song will have a constant BPM and you can shortcut this step a bit but that's the "long" way. At this point it's easy - plug that BPM into Stepmania and make the steps. Easy song. Life is good. :D

or

2) I verify that the music does NOT have a constant BPM. At this point I get an accurate estimate of the BPM of the BEGINNING of the song and enter that into the .ssc file.
From here, it's literally trial and error. I insert BPM changes as needed in the Stepmania editor until the entire song is in sync. With many rock songs, especially classic rock, this task is an absolute NIGHTMARE as the performers aren't constant at all. The "Real" BPM of the music is changing constantly and matching the arrows to that is a very tedious process, sometimes requiring syncing a song one beat a time. Rather than trying to, which is way more time consuming than it's worth, I simply aim for a very close approximation via trial-and-error in the Stepmania editor. I make some "dummy" steps that I can sync to before making the final version of the steps. I then meticulously go through the song, verifying each section of the song carefully.

Each section is vigorously tested WITHOUT WEARING HEADPHONES. It's absolutely VITAL that I can hear the click of my keyboard press when doing this because I effectively treat my keyboard like an "assist tick". I watch where each arrow lands in the receptacle at the top of the screen as I play back the song. If the arrows are disappearing above the receptacle, I know my arrows are coming in early and I'll fix the BPM in that section as needed and vice-versa. Some songs are more frustrating than others but sooner or later you WILL get everything lined up nicely.

While it's not 100% perfect because it's not been matched one beat at a time by mathematics, it's close enough for all intents and purposes - the perfect window is nearly in the middle of each arrow and a great will trigger on roughly the same distance of being "early" or "late" on each arrow. It's not 100% perfect but it's 99.5% perfect and plenty good enough to AAA(A) the song by hitting keys in time with the music.

It's important to note here that Just because you can get a perfect on each note, that does NOT mean the song is in sync! Watch WHERE each arrow is disappearing. The perfect window is a WINDOW. Are you able to step later than usual and still get a perfect but stepping a microsecond late gives you a great? Please fix that! Focus on getting each beat landing as close to the middle of your perfect window as possible. This is importatant for 2 reasons:
1) Stepping only a fraction off the beat shouldn't punish the player unintentionally and
2) Some people are playing Stepmania with tighter timing windows than you. What's a PERFECT to you may be a a GREAT for them!
You can turn on ProTiming if that helps you, too, and watch for patterns in the results as you play.

Sometimes it's a little tricky because the performers aren't tightly in sync with each other. For example, a guitar hit might be INTENDED to be at the same time as a cymbal crash but they're slightly off because one of the performers messed up by a fraction of a second. In this case, I always sync my arrows to the DRUMS/PERCUSSION of the music because it's the rhythmic driving force of the song. When there's a descrepancy, aim for being in sync with your percussion section.

Trust your ears. Your keyboard clicks should be perfectly in time with the music. Purposely hit notes a tiny bit early or late to be able to more easily see what sort of "bias" there in the perfect window at that moment so you can make the proper BPM adjustments. In time and with practice, you'll be able to sync up any song you damn well want, no matter how awkward the BPM is. :)

I'll post up one of the SM songs I've made steps to:
Bryan Adams - Summer of 69 (disregard the filenames of the .sm and .ssc files. It's an inside joke on my end. :P)
Note that the offset will probably be wrong for your system. I synched it on a system with a different global offset so it doesn't translate very nicely to other machines without manual offset tweaking.

Open this in the Stepmania song editor and note the large number of BPM changes that makes this possible. Notice that, while it's not mathematically 100% in sync, it's synched by ear closely enough than it's possible to play and enjoy without having serious sync problems. I could go back and tighten it up a bit more if I really cared to but honestly, it's good enough.
Go ahead and try to improve the sync of this song a little bit if you want some practice. It's already pretty close so it's a good starting point. :)
Once you see how much work goes into syncing a song with constantly-changing BPMs, you'll quickly start realizing that accepting "really close" is good enough. You won't get it 100% without more effort than basically anyone would be willing to put in.

I don't even know how you COULD reliably get a song like this 100% in sync guaranteed. There aren't always clear beats visible in a wave editor, the BPM changes every instant because of the nature of live performance and every instrument has its own unique BPM because, again, live players. I consider that an impossible goal so I simply aim for "good enough to AAA" while keeping the bias of the perfect windows in mind and tweaking as best as I reasonably can.

Last edited: 27 June 2015 2:22am

Reply
Or you could just use a much more dreadfully simpler and mathematical approach instead of using something as unreliable as ears (which will never be accurate enough to detect millisecond differences) or timing (that more often than not drifts a ton even in the tightest judgement window)


" that means that either your God has abandoned you or your song is just drifty by nature. "

I don't consider that acceptable. Saying "god hates you, deal with it" is a cop out. Fix it or make a different song! Songs that have the BPM wobbling all over the place relative to the music are obnoxious to play. When I find one in my song library, I delete it immediately. I know how much work they are to fix and I can't be bothered. When I take on a constantly-changing BPM song as a stepfile, it's a PROJECT. They take probably 10-20x longer to make than stepfiles with constant BPMS and 95% of that time is just piddly work getting each measure to line up just right. I see why a lot of people don't do it - it's time-consuming, hard work! It's the only option if you're going to make a song with constantly-changing BPMs though. Do it right or do a different song.

Fixing the sync of a drifty song requires compensation for the driftyness. I'm tired of playing stepfiles that average out sections of the music and multiply by the number of beats. That doesn't work. See the troubleshooting:


"You can still potentially sync your file to be fairly FA-able; you would do this by inserting very tiny bpm changes every couple of measures. (See my Battle 2 simfile in Mute Sims 7 for a good example of this practice in action."

That's basically exactly what I suggested doing. I'm just saying there's no nice way to deal with a lot of driftyness. You just need to fix it by ear, one measure at a time. There's no rocket science to syncing a song with constant BPMs, even if there are half a dozen of them within that song. Those kinds of songs aren't the problem. I agree - there's no excuse for messing those up. The problems arise when a song is constantly changing BPM and how a stepmaker can deal with that. I was simply describing how I personally deal with that.

Last edited: 27 June 2015 2:27am

Reply
Thank you guys. Mostly I was just wondering if there was a program that you guys used that showed the changes or not. whenever I get time, I'll play around with the timing and such. I'll hopefully get it correct xD.

Thank You Guys again

-Tal-
Reply
Thank you guys. Mostly I was just wondering if there was a program that you guys used that showed the changes or not.
Thank You Guys again

-Tal-


I find the best tool is simply a metronome + my ears.
For example, if I know that some of the song is a constant BPM (which I can also determine using a metronome), I simply set my metronome to that BPM and play back the music. If the BPM of the music changes, I'll IMMEDIATELY know exactly where that BPM change is, even if I can't hear the BPM change directly because it's so small. The metronome will tell me where it is and roughly by how much. From there I can tinker with the metronome to determine how MUCH of a BPM change there is.

The problem with using math as a tool is that it doesn't verify the BPM of the music at every given point like a metronome does. You need to be able to test the music against known BPMs so you can see where it changes.

Without a metronome, it can sometimes be hard to even tell that a BPM change exists at all. I recently worked on a song that was 190 BPM for most of the song but then a section went down to 188 BPM. I didn't know the song had a BPM change until I put it up against a metronome because the change from 190 to 188 was too small to really hear just by listening to the music. With a metronome going, I immediately discovered that a BPM change EXISTED, where the BPM change was and I had a tool to quickly determine the new BPM. It was relatively painless.

Rinse and repeat for the rest of the song and you're good!

Last edited: 29 June 2015 3:18pm

Reply
Any modern audio editor program (like REAPER/Sony Acid/Garageband [maybe audacity has it too now?]) gives you the option to put a beatmap of whatever bpm you want. You're literally just looking for a start of an instrument attack and placing the beatmap there and adjusting till it matches all the other attacks in the waveform. If for some reason it does stay constant for a while but change bpms, you're gonna see the beatmap there go off from the normal attack and you can simply adjust from there.

Metronomes are completely deceptive and limited to bpm changes quite large (above ~2 bpm discrepancies). If something drifts by say .050 millisconds, a metronome to your ears is not gonna show that, whereas on the waveform it's much more obvious.

Work in Progress
Reply
Any modern audio editor program (like REAPER/Sony Acid/Garageband [maybe audacity has it too now?]) gives you the option to put a beatmap of whatever bpm you want. You're literally just looking for a start of an instrument attack and placing the beatmap there and adjusting till it matches all the other attacks in the waveform. If for some reason it does stay constant for a while but change bpms, you're gonna see the beatmap there go off from the normal attack and you can simply adjust from there.

Metronomes are completely deceptive and limited to bpm changes quite large (above ~2 bpm discrepancies). If something drifts by say .050 millisconds, a metronome to your ears is not gonna show that, whereas on the waveform it's much more obvious.


I'd be interested in trying such a method sometime.
My first thought when I read your post is the challenge presented by busy music/waveforms, especially when accents in the music aren't landing on the beat. In such situations, accurately positioning a beatmap beat exactly on a beat would need to be deduced by evenly dividing the nearest "known" beats, would it not, because they wouldn't be detectable visually?

Here's an example of a song that might run into problems like that:
Nightwish - Fantasmic

The goal of the method you described is to get 100%, sub-millisecond accuracy at all times. I'm just wondering how that's possible if not every beat is clearly visible. Would you not have to settle with "close enough" simply because it becomes impossible to determine exactly where every beat starts, visually? As soon as you have to guess on one beat, all other numbers past that point will be speculations based on that guess. If the guess is wrong, everything past that point is wrong and the sub-millisecond accuracy is gone (if you change your guess, everything after that will need to be compensated for). Chances are, any piece of music with a variable BPM will have more than 1 "guess" scenario like this and the errors will quickly add up.

Last edited: 30 June 2015 4:37pm

Reply