Feels really in character for a Siivagunner joke to say that the one VGM song that nobody can agree on its time signature about was actually a 4/4 piece. Makes sense that they'd be the ones to know about it, too, considering they'd have had to actually try and fit this onto a timeline.
@@shytendeakatamanoir9740me when I tell you about SiIvagunner's other various jokes which aren't just Grand Dad like Snow Halation, The Nutshack, and more recently, Raft Ride:
@@user-AADZ me when I explain why the joke wouldn't have worked with more references like I am Joke Explainer 7000 (I was gonna mention Snow Halation, but it wouldn't have worked as well imo. It would have broken the rhythm, and made it less effective overall.)
I DM'd Shogo Sakai on Twitter asking about the meter of Strong One recently, and he told me he was thinking of Strong One in terms of *additive* time signatures: 3/4 + 3/8 + 1/4 + 1.5/4 (spicy!) It makes sense when you're focusing on the pulses in the combos driving the time signatures by themselves, but it also affirms the idea that Shogo Sakai *was* thinking about decently "regular"/"clean" meters, and therefore 15/8 and 29/16 kinda fit those conventional vibes. For even more context by the way, the mp2k/Sappy sound engine -- the engine sent out to GBA devs by Nintendo which was used in Gen. III Pokémon *and* MOTHER 3 -- has a tempo "resolution" of 2 BPM, meaning the next smallest tempo change above 120 BPM Sappy supports is 122 BPM, *not* 121 BPM. Also, excellent job accounting for the 3:5 16ths tuplet at the end of (Masked Man) ❤
wait a second this completely tracks, masked man version just cuts off the last note so 1.5/4 this means that strong one masked man is just in 7/4 and gets an extra 16th note from subpar tempo resolution
3:42 "It turns the otherwise familiar tune into a counterintuitive abomination." No spoilers, but that is just... saddeningly appropriate given the Masked Man's circumstances.
@arn3107 massive spoilers for the game's ending: So, at the end of the game, it's revealed that the Masked Man is Claus, Lucas' brother, who was presumed dead alongside their mother at the start of the game. Porky, the game's villain, found his body, added cybernetics, and stripped him of his free will, even going so far as to say that, "Claus almost sounds like a person's name; but this isn't a person anymore, it's my toy." Admittedly, this isn't the biggest shock in gaming history, but it's still a pretty big plot point.
@@littlebigb5370 thanks i actually already watched a video on Mother 3 so i already know the plot though i wasn't sure if what happened to Claus was Porky's doing but thanks for confirming it have a nice day
I think the whole issue is that we're trying to notate this song with a time signature even though the real, true intent of this song is to make drummers take confusion damage. Yknow, since following the rhythm that's literally the gameplay mechanic, it forces the encounter to be one against an opponent that you can't get a solid read on. I think the REAL most accurate way to notate this song is to put a note in front of the drummer section that says "have fun LOL". And put question marks behind the time sig. If a notation of this song doesn't appear on the "threatening music notation" twitter, then it's not accurate enough.
Also according to some other comments, Shogo Sakai *did* claim to actually think in terms of additive time signatures when composing this. Just because something is intended to be weird or jarring doesn't mean we should just assume it defies explanation. That's kind of a cop out
@@pedrogarcia8706 Its more a showcase on how time-signatures are not real. Meter is a concept of the brain to make sense of rhythm. (Hugo Riemann: System der musikalischen Rhythmik und Metrik. 1903)
I like how you say "verified by the official earthbound wiki" when it's just some fandom page lol makes sense silvagunner's wiki would know since it's actually full of musicians
@@aaronkandlik Not quite - as I've seen in an explanation, it's mostly because at that point those dudes had hung around and played together for so long they could intuitively get that bizarre, amorphous break.
what I love about Siivagunner is that they're kinda like IRL Mr. Saturns - they're weird, they're goofy, but damnit they know what they're doing. of course they would be the ones to figure out the actual time signature of Masked Man Strong One
So, interesting math note: You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3. You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain. In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8. The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song. In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536. In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576. Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go. (p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
@@Leekodot15 Yes! Traditionally the bottom number of a time signature will be a power of two, so for these I picked the next highest power of two after each number. I could have picked any power of two, but the closer the two numbers of a time signature are to one another the fewer extremely short or extremely long note lengths you need to notate And computer storage is also fond of the powers of 2, given that each bit you add to something multiplies it’s number of potential values by 2
13:17 it could be that the game actually did account for the timing error, and corrected it, but then was ran on an emulator that had the correct fps or higher, causing the song to play faster.
mother 3's battle system is so cool, the complexity of the music being tied to the difficulty of the battle is so creative and it helps that shogo Sakai is a mad genius with making creative vgm
the funny thing I only realized about the unique music mechanic.... after I've beaten the game, I didn't know I was playing on hard mode for the entire time.
Game dev here. My explanation for the weirdness you're seeing: Games run at 60 FPS or 30 FPS or whatever, but audio doesn't follow frames. It's handled separately, and constantly gets out of sync with the frames. (I'm making a rhythm game and MY GOODNESS it causes so much trouble.) Recordings of MOTHER 3 will have imperfect timing, but never enough for anyone to detect unless they check - like you did. For gameplay purposes though, it's accurate enough. Based on your video, it's a safe bet the heartbeat is handled automatically by the code based ONLY on tempo - which explains the quarter notes. Shogo Sakai possibly wrote these songs at 15/8 or 29/16... but converted them afterwards to be 4/4 with appropriate tempo changes. I bet MOTHER 3's team wrote a tool specifically for converting all songs to 4/4, to save time. Which makes sense: This way, every time they revised the music during development, the heartbeat wouldn't need to be manually corrected to match - it's already part of the song! This would explain why all MOTHER 3 recordings are slightly out of sync, and why technically all battle themes are in 4/4 time signature. It was probably the solution that was the least trouble. Hopefully you didn't pull out too much hair trying to decipher the time signature! There's always multiple ways to solve a problem when making a game; sometimes you just pick one that's good enough.
Scrolled down to say basically the same thing, and well put. The reason this isn't pointed out more often is because the difference that audio processing causes is often so small that it's negligent and goes completely undetected.
@@chupathingy13exactly. Supposedly humans can't detect things faster than 13ms, and a 60FPS game means the frames are around 16-17ms. But when it comes to audio, we absolutely do hear a stutter, Even if it's a lot smaller and just a little pop or crack. So usually they separate audio from frame processing. Fun fact, if you've ever had a piece of software crash, and it keeps making the last noise it made, I think this is why. The audio processor is still going, but it's not getting instructions from the rest of the software on what sound/note to play next.
@@kantackistancan confirm from knowledge of nintendo console architecture, this is all exactly it. The GBA streams raw 8-bit PCM audio from the CPU most of the time, unless it's using gameboy emulation features. So all audio playback is at the mercy of the CPU's timing for game code. It's possible, if not likely, that two recordings from console with different gameplay states wouldn't even sync up 100% over time. With little microseconds of lag every time a button press is registered and it has to branch to its process.
I’ll note that on older hardware this was even more suspect because frequently the audio chip and the cpu were not so separate. Super Metroid on the snes for example offloaded some processing to the audio chip, and thus doors can only open when sounds have finished. This actually made its way into the tool assisted speedrun for super metroid, as a significant consideration for optimisation. I assume the GBA had a clearer distinction between audio and game processing, but it’s totally possible that the audio chip runs slower the harder the cpu works.
It's also possible if not likely that recordings from 2 different GBAs would have slightly different playback speeds. The crystal oscillators that all signal timings derive from are intended for consumer devices, not scientific applications, and have some tolerance for deviation from nominal speed; I wouldn't be surprised if discrepancies of ±0.01% between units at identical temperatures are observable.
The way it's programmed to handle the heartbeats is fascinating to me, I know that several other games have sound engines on the GBA that use MIDI files in this way (hello everything at Game Freak), so it's unsurprising that they'd use them for this. But I didn't expect the *hard coded* use of quarter notes, that Shogo Sakai and/or a sound engineer assisting him would then have to convert into something the game could then use for its own system. The fact that this also makes the nightmare of Masked Man into 4/4 makes me wonder if Sakai *was* also the sound engineer at the same time he was composer (which would track, as a lot of older soundchip hardware for NES and SNES had the same thing going on) because this is too elegant to be handled by two individual people for how monstrous it is to our ears. I don't know Mother 3's internals, but as a programmer... I think it's hardcoded and less of a limitation, as being able to effectively listen in for different note types isn't theoretically hard to do if you have access to a full MIDI track on actual hardware. It's more of a "build to scale" thing - Mother 3 has a lot of tracks, like over 200. Not all of them are used in battle, but many of them are or can be at certain times, so your beat tracking system needs to be robust to account for all of them. It was probably just a shortcut and the creation of a 0 volume MIDI track with exclusively quarter notes appended to the end of each bit of music data, that every song then needed to be engineered around after the fact. It's for a similar reason that all the instruments used by the Bash command are in C minor - though that also has both lore reasons and limitations on what kind of battle themes could be composed. Sakai is a wizard. I think Strong One's variants are supposed to *sound* like 29/16 and 15/8 respectively, and I agree with the conclusion that we can treat them as "approximants" of these time signatures. Less because of the limitations of the GBA, but more because that was the artistic intent to be "just off" of them to make them harder to follow! After all, if we approached Dry Guys *not* as 3:2, musically that makes no sense, even if the game's actual data speaks otherwise. These slight tweaks are intentional because of Strong One's purpose - if Sakai wanted to make them 29/16 or 15/8 as opposed to...whatever this tempo map is, he could've, clearly. (I'm not in the man's head, but yeah lol) He wanted to mess with us a bit :) ...WAIT HUH LOL THE REFRESH RATE JUMPSCARE Great overview, this was a joy to watch!! EDIT: Thanks for the like :) Saw a typo I had to fix but UA-cam gets rid of it on edited comments, ah well
Someone in another comment said that the GBA Pokémon games use the same sound engine as MOTHER 3, which would explain why the heartbeat 'structure' is so similar.
I cannot *believe* the thing about the heartbeat tracks all being 4/4 quarter notes and hacking together anything else with tempo changes. What the hell.
I love how everytime you talk about something in music theory you then add that to your outro music because it makes me realize that without you explaining it I am STUMPED
9:59 Imagine you're a composer writing for a videogame and suddenly you have to do weird time signature, speed and rythm calculations so that it's percieved as written above
71907/40120 is mathematically equivalent to the 4/4 beat changes. With 71907 beats in a bar, heartbeats 1,2,3 and 7 get 10,030 beats each, heartbeats 4 and 5 get 7,021 beats each, heartbeat 6 gets 5,355 beats, and heartbeat 8 (which gets divided into a triplet) gets 12,390 beats. It doesn't actually matter what the beat denomination is, but if you choose 71907/16384, the tempo (whole note) is the same BPM as the framerate in FPS. I don't know what the other math guy is thinking.
Taken from another comment: So, interesting math note: You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3. You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain. In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8. The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song. In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536. In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576. Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go. (p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
I chuckled a bit when I heard “verified by the official earthbound wiki.” =P It’s a fan-run wiki, so it really has no authority to verify anything. They just curate information that’s believed or discovered to be true by superfans.
I mean, for stuff like this, that's about as close as you can get, especially if a source/proof is provided. At the end of the day, any sort of "official" verification depends on trust of the supposed authority.
Well...if you want to really break the term "verified" down like that, no authority on Earth can truly "verify" anything. I've found the Earthbound/Mother community to be very passionate about the series, so if any low-level discoveries were to have been faked or misreported, the community would pounce on it. I'm sure you could do your own investigation into the matter and see whether their findings line up with the facts. Besides, not to bring politics into the discussion, but the US's CDC during the COVID pandemic made very bold "verified" claims that turned out to be false or made up on a whim, such as the 6-feet social distance rule being allegedly necessary and based on studies.
@ what I’m saying is that for information regarding a video game to be verified, it should come from a first hand source of some kind. For example, fans can speculate about the Zelda timeline, but if Nintendo, Aonuma, or a licensed publisher working directly with Nintendo offer up information about the Zelda timeline that coincides with that information, the information is now verified.
5:52 Hey, it's me! I added those a long time ago. Quick note about the wiki, it's not official by Nintendo, and even then the wiki and its community have abandoned Fandom ages ago and are now an independent place, WikiBound. Years ago I was playing with the game's memory addresses and found the address for the current BPM -- I have a video showcasing the value and what happens if you change it. That is where I got the BPM values I added to the wiki from. I do however remember that they never exactly matched real life; I think I vaguely remember checking that Zombeat is internally at plain old 120 BPM but when you actually hear the song the BPM is slightly different. Most likely because of the GBA's clock like you said, but also misc. stuff like limited decimal place calculation capabilities and lag frames. I have not looked into how the game works very deeply, but I did gather some conclusions back then, and have made use of them to developed a since-abandoned project, PK Rhythm, which mimics Mother 3's combo system. In both Mother 3 and PK Rhythm, the system is pretty much what you describe: whoever made the song specifies BPM changes and then the game just checks if the player's hit is on-beat or not. Both are completely ignorant to the concept of time signatures, off-beats, etc.
Cadence, as a former musician (I say former because I never get to gig or teach anymore), this is exactly the kind of content that me and so many other aspiring music professionals wanted to make and watch when we were younger. The fact that you are actively researching, making content, and experimenting with music so earnestly means that you are on your way to great success as a content creator and musician. Keep improving, keep learning, and keep writing. Great video!
Oh my God this is one of those times where you come across a video and feel like it was made specifically for you and you alone. This was fascinating, thank you so much
Indie game dev here. Chances are the reason why it's not exact is because you have a variable amount of time that each frame is being processed before it gets to queueing up music. Let's say Frame 1 it starts immediately, then 30 frames later it's time to queue up the next bit of music. Before it gets to the music, it may be processing some other chunk of data such as player input that takes some milliseconds. So then the music gets queued slightly later. If the code is looking at the projected end of a string of notes, the next sections all get pushed out slightly each time this happens.
For those wondering where the 71907/40120 time signature came from: In the actual sequence data, the tempo is not consistant at all. The tempos are: 3/4 at 126 bpm 2/4 at 180 bpm 1/4 at 236 bpm 1/4 at 126 bpm 1/4 at 102 bpm The total length of time for this sequence, is precisely 23969/7021 seconds (that's the simplified fraction). It should be assumed that the intended tempo of Masked Man is 126 bpm. Now, if something is playing at 126 quarter notes per minute, and each measure is 23969/7021 seconds, how many quarter notes is this? At 126 per minute, each quarter note lasts precisely 60/126 seconds, which simplifies to 10/21 seconds. Now if we divide 23969/7021 by 60/126, we get 3020094/421260 quarter notes. This simplifies to 71907/10030 quarter notes. Dividing 71907/10030 by 4 gives us 71907/40120. Even the regular Strong One isn't _exactly_ 15/8. It's something like 14.88/8.
Taken from another comment: So, interesting math note: You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3. You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain. In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8. The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song. In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536. In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576. Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go. (p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
I'm not convinced that the interpretation of "it's 4/4, just speeding up and slowing down" is a useful one from a music theory perspective. Sure, that's how it's implemented in-game, but as you point out with Dry Guys and all the other tracks, you can really do any beat that way, even ones that are obviously intended to be in a totally different metre. That said, the fact that the ratios make it a good distance from 29/16 is pretty convincing. But you also raise a good point at 11:15 in that it MIGHT have been meant to be 29/16, although the extent to which it's off makes me wonder if it was actually meant to be something different even from that. It does make me wonder what might be a good compromise between the somewhat-off 29/16 and the patently-ridiculous-but-technically-correct 71907/40120 I've seen floating around.
Taken from another comment: So, interesting math note: You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3. You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain. In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8. The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song. In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536. In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576. Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go. (p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
Sounds to me like Strong One was written in 15/8 but as they were putting it into the game somebody messed up the tempos, and they thought "hey wait that kinda sounds cool, what if we use that for a harder enemy encounter?" The best creative decisions are often the ones found completely by accident.
Composer: Ho ho, this one's going to throw people off. They think it will be in 7/8, but it's actually in 15/16! Intern: Actually sir, your calculations are a little off. It's actually a 43875x10^15/64 Time Signature Composer: Ah it's okay. It's not like anyone will notice 😅
Having ripped a fair amount of MIDI data from games, I can tell you that much of the time a time signature isn't even programmed and it's technically 4/4, even if the music itself is just in 3/4 all the way through. I also actually saw someone point this out about "Strong One (Masked Man)" somewhere, but people just got mad and insisted "well it's still effectively 29/16", which this video demonstrates is false. Side note: that is not the "official EarthBound wiki" and just a fansite hosted on Fandom (blech). But the SiIvaGunner wiki is edited by people associated with the channel, so that one could possibly be considered official- for SiIvaGunner of course, not EarthBound.
The SiIva wiki is mostly handled by fans of the channel, though with how big the SiIvaGunner team is there ends up being a decent amount of contributors that also write stuff for it and are involved with managing the wiki.
i think the reason why the heartbeat notes are straight quarter notes with changing tempo is so that the combo sounds of the characters will flow seamlessly-ish into the next combo sound
A decent idea as to "why the in-game timing is slightly off from the expected ~59.7hz" is simply "its lagging a little". The GBA didn't have a dedicated sound OR graphics chip, so _everything_ went through the CPU. Graphics, Audio, Game Logic, you name it. Understandably, this is going to cause a little (edit: basically imperceptible to the human senses) slowdown - especially as the GBA gets older and hardware starts slowing due to age. Unless you find a video someone took of their GBA in 2006, and even then to a lesser extent, there's going to be inaccuracy.
What’s insane is that even though the timing in this song is so strange after enough listening the groove starts to make sense! Repetition legitimizes, repetition legitimizes
This helps really illustrate the function of tume signatures. Its so easy to just wonder "why?" as a person who only ever dabbled in music and who was good at learning by ear. I mean, you show how most people are wrong, but you show how you can be so precise and represent something that feels like ear chaos succinctly on paper and for a machine with a purpose, which is cool
Idk what you're talking about disappointed, this song being in 4/4 with wild tempo changes is such a great rug pull, and also much less common in the context of avant-garde music, and ALSO more nicely responsive to the affordances of computer-played music. Loved this video to bits! Thank you!
i remember finding out about the battle songs being 4/4 with different tempos when i dumped the midis for the songs when i wanted to remix them, it was super interesting to hear fate/serious as it was programmed and i remember i talked about it in a random youtube commend like 7 years ago (will never ever find it again though). that said i don't really think it's very useful to just say "in the end, it's 4/4 but with tempo changes!" (even if it's technically correct) especially considering that applies to literally every battle song in the game. the datamining of the music files and seeing how the tempo behaves surely gives some very interesting insight and it would have helped here, had the game not had that rhythm mechanic. like some other commenter said, shogo sakai clearly had a time signature in mind and i think creator intent should be what matters here. hell, for all we know it might have been his intention to make the song not fit in any steady tempo just to throw the player off even harder (after all it's considered to be notoriously harder to combo on than regular strong one). if things don't sync up, especially in an old console, that can just be attributed to hardware inaccuracies, just as you said; after all you'll find that even NES games had trouble keeping a consistent tempo at all let alone a tidy whole number one. the fact the song seems to sync up really well at 29/16 at first tells me that 29/16 is a perfectly valid time signature to classify this song and i don't think it's worth diving into why it doesn't *perfectly* line up. lastly i have a feeling that the programmers simply felt like it was way easier for them to program the heartbeats to be in sync with the tempo and just told shogo "hey you'll have to play with the tempo lol sorry" who knows they probably wrote a script for him that generated tempos for him so he didn't have to manually calculate the ratios, but i don't think it was a hardware "limitation" but rather a "this is the easiest thing to code lol". tl;dr for me this song is still 29/16 and if it doesn't line up exactly at that it can just be attributed to hardware and it's not really that important. p.s. 5:14 fucking killed me you're hilarious
Taken from another comment: So, interesting math note: You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3. You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain. In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8. The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song. In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536. In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576. Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go. (p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
Well, it could be said that time signatures in general tend to be approximate anyway. Classical music notation was never meant to exactly match the performance (nor the data from the game in this case), and neither using time measures (not talking about the 29:16 though, as it's still kinda manageable) nor precise BPM changes that are impossible to intuit feels satisfactory for a transcript, as like you said - it isn't helpful for performance - but I'd go further and say it isn't helpful for musical analysis, nor for any case in which the staff notation may be needed. I'd treat it similarly to other stuff that such notation doesn't usually notate (and even if so - never precisely), such as timbre or other things that appear only in a recording
That said I'm not a fan of notating it as 29:16 either as it still feels like there is a tempo change in the loop, but I'd probably just eliminate all but one of them by using approximations of the BPM ratio, possibly having a relevant symbol or annotation that the note length is a bit shorter/longer
Definitely true in a lot of cases, especially performance and stuff like you said. I think it's important in this context for two reasons: The tempos/time signatures are programmed in, and there was some level of deliberation in the exactness of timing that I think is worth analyzing, i.e. the feel if dotted eights:quarters is 4.29:3, made to deliberately feel like you're rushing. Also in terms of sync, any mashups or covers of this tune that intend to use the same beatmap need to know the temple cycle or it will desync.
Love this answer! Only thing is I would still argue 29/16 is sensible since the final eighth note in Strong One is effectively "cut in half"; it's the most straightforward musical representation of what Shogo Sakai did with the song's meter, IMO.
It's so cool how different representations in sheet music can give the same sounds! It's also neat how the music is *programmed* can end up very different from how one would write it for, say, actually making it playable for a live band, or comprehensible to laymen otherwise. Like, I doubt you'll have more people able to mentally conjure how rapid sudden tempo changes sounds compared to weird time signatures and odd note notation.
You can approximately count it as "quarter, quarter, quarter, triplet, triplet, eighth, quarter, quintuplet quintuplet quintuplet" at 123.6 bpm, and it should loop nearly perfectly with the worst note being only 1/50 of a second out of place. In this interpretation, the time signature would be 211/120 (as 3/4 + 2/6 + 3/8 + 3/10), or ~7.033/4, which is only so complex because of the unresolved tuplets.
Taken from another comment: So, interesting math note: You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3. You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain. In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8. The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song. In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536. In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576. Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go. (p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
Honestly this is such a great video. I know nothing about music stuff but I’m a huge Mother 3 fan and I’ve always thought something wasn’t quite right about that song.
Well I'll be damned! I was a little salty to not hear you talking about this in your "weird time signatures in Nintendo Games", but with this whole video dedicated to this specific song, I'm more than happy and, more importantly, completely wrong in my comment on your previous video! I'm sorry for the strife and thank you for making this video! I absolutely love mother 3 and hearing you walk through the technicalities of this song was a blast, please keep up your amazing work!
I cannot express in words how even though I’ve been into music with weird time signatures my whole life, I experienced severe psychic damage after hearing the drum part to this. I can’t even imagine it would sound good played by a real drummer
If you think of sheet music as a representation of ideas to guide musicians on how to play a piece (from the days before audio recording) then the actual music that comes out is the real source of truth for how it sounds, and that will vary every single time the piece is ever played. For example listen to any rock music with drums that wasn't recorded to a click and the tempo will fluctuate constantly throughout the song, but you wouldn't notate out the tempo change on every single bar (unless you have way too much free time) because it isn't useful to the reader trying to play along to the sheet music - you would probably just write "120bpm (played loosely)" or something like that Computer music (particularly on older consoles) is in such a weird place because you have to define in strict mathematical form EXACTLY what is to be played and the computer will strictly interpret that using whatever method it has The song would be notated for humans as 29/16, it would be notated for the gameboy engine in 4/4 with tempo changes, and in reality it is whatever the gameboy decides it is when it processes that information (which might technically vary on each play through, on each different gameboy, each time the temperature of the circuit board changes, etc), which again is different to what we would hear on a PC/Phone after ripping it from the gameboy TL;DR - Reality is fake and we live in a simulation
Taken from another comment: So, interesting math note: You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3. You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain. In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8. The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song. In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536. In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576. Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go. (p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
Every time I hear someone talking about time signatures, at least as they relate to video games, I am brought back 15 years to a top 10 list from GameFAQs, top video game songs in weird time signatures. The number one place was attributed to Tor, the final boss theme of Iji (one of the greatest, most underappreciated indie games of all time), being in something like 27/16. I may not remember it exactly, but its something I have been haunted by these last 15 years.
I feel vindicated. There was a period of several days where I would spend 30 mins-1 hr listening to this piece at a time, trying to figure out its time signature in light of inconsistant claims, and the conclusion I reached by listening to the music was that it was in either 4/4 or 8/4. 4/4 seems correct to me - there is more emphasis on the fifth note of the tempo cycle than I would expect in 8/4 (beats 1 and 5 seem equal in this piece, where they wouldn't be in 8/4), but I can see the argument for 8/4, being that each measure now contains one full tempo cycle & it is no longer awkwardly split across every pair of measures. Thank you for making this video. Hopefully this will set the record straight on how this piece works, in a musical sense. (minor edit improving readability)
12:00 well, sound emulation will never quite be 100% fully accurate so if you want the real ratio then there's no other solution than plugging an AUX cable in the headphone jack of a GBA and capturing the audio yourself otherwise, you're going to have discrepancies. I have this problem all the time when trying to fit SNES music to the grid in a DAW, it just always drifts
If anyone has watched this video and is now super interested in Sakai-san's music, it's worth noting that a few days ago he released a new song, "Necos Calm Me," on streaming! Also, that drum fill is NASTY.
As a computer science major, I'd say this may be a case of how time is calculated in the game's programming. To my understanding, some games in the past counted cpu clock cycles to time things right. If you go back far enough, you see games where the programmer had to literally calculate how long each instruction would take to execute, which is why many games in the 80s lacked background music during gameplay (think Galaga and pacman), because you can't predict what instructions specifically will execute, and thus can't time out music. Now, later on, counting clock cycles became a subprocess ran by the hardware itself, and all the programmer would have to do is check how many cycles have passed. However, there remains two problems: 1. The cpu clock on a Gameboy advance only has to be consistent enough to execute instructions. Beyond that, there's no need to make sure all manufactured cpu clocks run at precisely the same speed, which could explain why your recorded audio is different. 2. The point in the code at which the programmer checks the number of clock cycles may differ for the same reason as before: whatever actions the player takes may cause the check to occur earlier or later in the game loop. Now, in modern computers, we have much more precise clocks, better methods of checking cycles, and can even synchronize with remote atomic clocks. But back in the day, that kind of effort just isn't worth it when the yield is a perfect time signature, especially when the bottleneck is fitting the music in the cartridge itself. So I think your last theory was most likely correct, and you can chalk the unsyncronized nature of your last recording up to either programmer or manufacturer error, not musician. Great video btw! Edit: I almost forgot, many games synchronize the gameplay to the display refresh rate. Same principle, except instead of counting clock cycles, we're counting frames.
I think the main reason the song in game doesn't line up 100% with the midi data is purely due to: 1. The GBA having inconsistent framerate count with audio. (This is stated in the video) 2. The game may either purposely or unintentionally slow down the song because it has to change the tempo frequently, but it must also keep the heartbeat on track to make sure the timing doesn't desync as the battle progresses. It seems hard enough to keep track of the songs time signature in one loop, but in one battle it may loop several times. Therefore, there is likely something in place to ensure that the heartbeat doesn't loose track of the song itself.
it makes sense that the heartbeats would always be quarter notes. your heart speeds up and slows down - it doesn't play dotted notes. bpm changes are how our hearts work.
as a cs major, your first mistake is assuming that the gba runs at a 100% consistent framerate. Most computers will fluxuate a lot in their performance. And while the average is 59.90089 whatever, the gba needs to compensate for its lower bound of performance. In short, the music is likely truncating that 59.90089 acerage down to just 59, and playing the music at a tempo that will work for that fps (hence 0.58% slowdown)
I'm gonna be honest, I actually hear it in 7/4, or really a 2 measure grouping of 3/4 + 4/4. It could be an additive time signature but it'd be irritating to notate (and read) so I'd literally just give it the time signature (3/4 + 4/4) (in parens and everything) and alternate meters without putting another time signature down. It just has a hemiola-like effect in the 2nd bar. First measure is just 3 quarters. 2nd measure in the group is: dotted-8th, 16th tied to 8th, 8th tied to quarter, triplet 8ths. There's no rubato. There's no crazy time signature. It just feels weird because it's obscuring the beat in the 2nd measure of the groupings. There's nothing to hold onto until that last triplet. I even notated it the way we'd typically have done it in publishing to not *visually* obscure the beat if it were notated.
The timings of those rhythms are off with what is actually happening. It's a decent approximation but it very quickly desyncs, as I demonstrate in the vid. The actual tempos were datamined from the actual game files and the ratios are calculably different from the 29/16 (or 7/4) approximations.
You are correct that, at least in this case, it simply used the quarter note for the entire piece. In order to facilitate the metric changes it adjusted the tempo. We did this a lot in early notation software when it was difficult to change the note-value that received the beat (for playback or rudimentary mixing), and I suspect this is similar (or literally the same) to how MIDI is programmed. That being said I don't fully understand the nitty-gritty of MIDI programming, but I suspect it was due to the available resolution of the MIDI timing. If you're looking at the game data files, I'll have to believe you on that. You don't actually show that (the MIDI file) in the video. You're not importing the MIDI into notation software are you? Just looking at how it notated rhythms it seems like you might have? Quantization settings can mess with that, and older MIDI and notation software were notorious for not working together well. If this is what you did I would strongly suggest you rethink. I worked on some film scores having to import MIDI and you frequently had *slightly off* meters like this, because the software quantization needed tweaking to get a best-fit. Heck I saw that all the time trying to import MIDI. I'm still unsure about the time changes you get from the MIDI. However, you truly never know what can happen with MIDI, as it can be performed live with a controller, and any notation-import is a translation. Sometimes a digital (exact) translation makes the least sense because the MIDI timing is, for notation software, comparatively continuous, not discrete. This is why you have to tweak that quantization. Traditional rhythmic/meter notation, which I believe this discussion is about, is symbolic and not continuous and if you took any performance and used a high level of granularity (e.g., milliseconds) in determining meters, rhythms, and tempos you'd get crazy stuff, not even counting expressive changes. I just always try to look for the simplest explanation. This discussion really gives me deja vu of MIDI quantization issues. Just listening to the track itself, I'm not sure desyncing is actually relevant to the notation, but I think you're correct it's some kind of incompatibility between the timing of the MIDI and the game/engine frame rate. You might be interested in MIDI beat clock and timecode regarding possible explanations here. Beat clock (more specifically PPQN/MIDI resolution) might even explain why the tempo changes might be necessary in the programming if I'm understanding it correctly. Perhaps I'm just used to interpreting clear-cut, rounded boundaries in meters (at least of non avant-garde stuff) and my ears are biased. I'm not a MIDI expert but I suspect that someone who is would be able to identify the incongruities.
I always thought that this was supposed to represent how the masked man is actually Lucas' brother. Seeing your own brother, even if you don't recognize him, turned into an emotionless cyborg is gonna throw you off.
Man I always wondered about this song when I was playing, it was so weirdly out of tune but somehow in-tune at the same time, so I never got the timing right Great video! Love to see some Mother 3 videos and that little reference to Rhythm Paradise at the end
Halfway though the vid, I was starting to sniff sound card and CPU clock weirdness. I'm really glad you also thought of that and addressed it because I wasn't expecting it haha
Ok, this is what I understood from this piece. I think the melody was thought of as two groups of three notes. In the first version, the first group is a triplet of quarter notes and the second group is three eighth notes. In the second version the two groups of three notes of the melody have been altered in tempo, leaving the fragments without melody unchanged. The first three notes can always be thought of as a triplet at 120 on the metronome. In fact, this makes it clearer that it is a very small change and is well conveyed by the small difference between 126 and 120. The second part at 126 "without melody" does not come in after the triplet has taken place regularly, but a a little earlier, giving the idea of two "overlapping pieces". After that, the third part at 102 can be seen as a very rushed version of the group of three eighth notes, in fact I would write it as 3 eighth notes at metronome 204, rather than as a triplet.
Edit: my theory is that the creator had that specific music in his head or had recorded It previously and just found some tricky way to aproxímate It with the software that they were using and make It sound similar to what they wanted and It somewhat worked Hey i just wanted to comment that your videos are super super interesting. I subscribed from another video but since then you published a few more and didnt disappoint at all!
did you ever consider irrational time signatures? it sounds very much like one to me. it would be exactly what you wrote at 3:14, except that the first six beats would consist of a bar of 6/4 (or 24/16), and the last three notes you wrote as "3:5 sixteenths" would actually be placed in a separate measure of 3/10 and consist of 3 eighth-notes. talking in the original meter, those three notes would be played at the speed of eighth-note quintuplets (just with only three of them instead of the expected five), but it would be cleanest to write them in the way i mentioned. if you really wanted to put it all in the same measure it would have to be 36/20
people saying "all music is 4/4 if you don't count it like a nerd" as if changing the tempo multiple times per bar isn't some advanced music nerd shit lmao
Shogo Sakai made a song so fucked up people are still debating its time signature nearly two decades later. That’s crazy
perfectly 200 likes, nobody like this comment anymore please
NOOOOOOOO ITS ACTUALLY 201
@@xaigamer3129 bros lucky number is 200💀
770 let's go!@xaigamer3129
@@Neo36563 :0 THANK YOU
I have to say, "4/4 with vomit inducing, horrible, puke-summoning Rubato" was not what I expected.
Rubato that gives you car-crash-grade whiplash
It was what I expected.
(I have played the drums for a couple of years, around 4 years ago and I've also played rythm games my whole life).
@@chrisheartman9263I thought I was an alright drummer until I played this game xD Most fun combat system ever
Now have 79/32 with vomit-inducing, horrible, puke-summoning Rubato
This makes me wonder, how would this sound with a consistent tempo?
the literal example of "every music is 4/4 if you don't count it like a nerd."
Does this mean "no music is 4/4 if you count it like a nerd"?
All music is 1/4. Every note just has a different tempo.
1/1 if you count it like an idiot
1/1 if you suck at counting
0/0 if
Feels really in character for a Siivagunner joke to say that the one VGM song that nobody can agree on its time signature about was actually a 4/4 piece.
Makes sense that they'd be the ones to know about it, too, considering they'd have had to actually try and fit this onto a timeline.
What if Silvagunner was right about everything
@@Birdsflight44Every single music is actually a modified version of Grandad
They do upload high quality tips after all
@@shytendeakatamanoir9740me when I tell you about SiIvagunner's other various jokes which aren't just Grand Dad like Snow Halation, The Nutshack, and more recently, Raft Ride:
@@user-AADZ me when I explain why the joke wouldn't have worked with more references like I am Joke Explainer 7000
(I was gonna mention Snow Halation, but it wouldn't have worked as well imo. It would have broken the rhythm, and made it less effective overall.)
YOU THOUGHT IT WAS 29/16, BUT IT WAS ME 4/4
“YOU WERE EXPECTING SOMEONE ELSE, BUT IT WAS ME DIO”
YOU THOUGHT IT WAS A TRANSLATION MOD, BUT IT WAS ME DOLPHIN VIRUS!
A JOJO REFERENCE?!?!?!
Kono 4/4 da!
KONO FORU FORU DA!
I DM'd Shogo Sakai on Twitter asking about the meter of Strong One recently, and he told me he was thinking of Strong One in terms of *additive* time signatures: 3/4 + 3/8 + 1/4 + 1.5/4 (spicy!)
It makes sense when you're focusing on the pulses in the combos driving the time signatures by themselves, but it also affirms the idea that Shogo Sakai *was* thinking about decently "regular"/"clean" meters, and therefore 15/8 and 29/16 kinda fit those conventional vibes.
For even more context by the way, the mp2k/Sappy sound engine -- the engine sent out to GBA devs by Nintendo which was used in Gen. III Pokémon *and* MOTHER 3 -- has a tempo "resolution" of 2 BPM, meaning the next smallest tempo change above 120 BPM Sappy supports is 122 BPM, *not* 121 BPM.
Also, excellent job accounting for the 3:5 16ths tuplet at the end of (Masked Man) ❤
The tempo resolution probably explains the desync.
interestingly, if you add together those signatures, the result is actually 14/8. maybe I just forgot how to get a consistent signature from multiple.
wait a second this completely tracks, masked man version just cuts off the last note so 1.5/4
this means that strong one masked man is just in 7/4 and gets an extra 16th note from subpar tempo resolution
I did not expect to see sorting algorithm man here
@@妛槞 You'd be surprised how much work I've done on MOTHER 3!
3:42 "It turns the otherwise familiar tune into a counterintuitive abomination."
No spoilers, but that is just... saddeningly appropriate given the Masked Man's circumstances.
can you explain
@arn3107 massive spoilers for the game's ending:
So, at the end of the game, it's revealed that the Masked Man is Claus, Lucas' brother, who was presumed dead alongside their mother at the start of the game. Porky, the game's villain, found his body, added cybernetics, and stripped him of his free will, even going so far as to say that, "Claus almost sounds like a person's name; but this isn't a person anymore, it's my toy."
Admittedly, this isn't the biggest shock in gaming history, but it's still a pretty big plot point.
@@littlebigb5370 thanks
i actually already watched a video on Mother 3 so i already know the plot
though i wasn't sure if what happened to Claus was Porky's doing
but thanks for confirming it
have a nice day
@@littlebigb5370sounds eerily familiar to the Naruto plot with Obito Uchiha
"no spoilers but, [spoilers]"
I think the whole issue is that we're trying to notate this song with a time signature even though the real, true intent of this song is to make drummers take confusion damage.
Yknow, since following the rhythm that's literally the gameplay mechanic, it forces the encounter to be one against an opponent that you can't get a solid read on.
I think the REAL most accurate way to notate this song is to put a note in front of the drummer section that says "have fun LOL". And put question marks behind the time sig.
If a notation of this song doesn't appear on the "threatening music notation" twitter, then it's not accurate enough.
Psychic Damage is only appropriate, considering the franchise!
Giving drummers psychic damage was not on my 2024 bingo card but frankly i think this is the highlight of the year lmao
Time Signature
Good
---
Luck
Doesn't mean we shouldn't try. It's an interesting exercise
Also according to some other comments, Shogo Sakai *did* claim to actually think in terms of additive time signatures when composing this. Just because something is intended to be weird or jarring doesn't mean we should just assume it defies explanation. That's kind of a cop out
This is even more proof of the joke “Everything is in 4/4 if you don’t count it like a nerd”
in this case it's more accurate to say it's in 4/4 if you DO count it like a nerd, it being the tempo.
@@pedrogarcia8706 Its more a showcase on how time-signatures are not real. Meter is a concept of the brain to make sense of rhythm. (Hugo Riemann: System der musikalischen Rhythmik und Metrik. 1903)
Or 3/4
doesnt sound like a joke to me, sounds like bs.
@@cupuacu4life13 You’ll never believe what most jokes are
I like how you say "verified by the official earthbound wiki" when it's just some fandom page lol
makes sense silvagunner's wiki would know since it's actually full of musicians
I probably should have checked the URL a tad closer lol
Well of course! He only uploads *HIGH QUALITY* video game rips!
The SiIvagunner community unironically packed with goatee musicians
Siivagunner team puts so much effort just to make shitposts and we LOVE it
@@user-AADZYou should check out their mustaches!
this is like how in Master of Puppets the entire song is in 4/4 but then there's that two hit measure that people describe as 21/32
That’s just because…Lars
@@aaronkandlikThe riff was written by James though
Exactly.
@@aaronkandlik Not quite - as I've seen in an explanation, it's mostly because at that point those dudes had hung around and played together for so long they could intuitively get that bizarre, amorphous break.
what I love about Siivagunner is that they're kinda like IRL Mr. Saturns - they're weird, they're goofy, but damnit they know what they're doing. of course they would be the ones to figure out the actual time signature of Masked Man Strong One
So, interesting math note:
You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3.
You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain.
In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8.
The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song.
In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536.
In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576.
Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go.
(p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
".../65536"
Wait, that's familiar...
*One google search later*
...That's the same number of integers 16 bits can hold!!
@@Leekodot15
Yes!
Traditionally the bottom number of a time signature will be a power of two, so for these I picked the next highest power of two after each number. I could have picked any power of two, but the closer the two numbers of a time signature are to one another the fewer extremely short or extremely long note lengths you need to notate
And computer storage is also fond of the powers of 2, given that each bit you add to something multiplies it’s number of potential values by 2
I got disgusted when i read “631890/1048576” Great math!
YES, THANK YOU
I agree with your comment 110%.
so tl:dr 29/16 is actually correct?
13:17 it could be that the game actually did account for the timing error, and corrected it, but then was ran on an emulator that had the correct fps or higher, causing the song to play faster.
mother 3's battle system is so cool, the complexity of the music being tied to the difficulty of the battle is so creative and it helps that shogo Sakai is a mad genius with making creative vgm
Gotta wonder about Earthbound 2 on the N64 though.
Still kinda want, in the same way OoT is friggin OoT, but these Zelda 64 scraps are fascinating...
the funny thing I only realized about the unique music mechanic.... after I've beaten the game, I didn't know I was playing on hard mode for the entire time.
As a drummer, the video just got more and more horrifying just imaginating having to play that. Very interesting tho !
Felt this as a tuba player. keeping the rhythm is hard enough in normal sane grounded music
Game dev here. My explanation for the weirdness you're seeing:
Games run at 60 FPS or 30 FPS or whatever, but audio doesn't follow frames. It's handled separately, and constantly gets out of sync with the frames. (I'm making a rhythm game and MY GOODNESS it causes so much trouble.) Recordings of MOTHER 3 will have imperfect timing, but never enough for anyone to detect unless they check - like you did. For gameplay purposes though, it's accurate enough.
Based on your video, it's a safe bet the heartbeat is handled automatically by the code based ONLY on tempo - which explains the quarter notes. Shogo Sakai possibly wrote these songs at 15/8 or 29/16... but converted them afterwards to be 4/4 with appropriate tempo changes. I bet MOTHER 3's team wrote a tool specifically for converting all songs to 4/4, to save time. Which makes sense: This way, every time they revised the music during development, the heartbeat wouldn't need to be manually corrected to match - it's already part of the song!
This would explain why all MOTHER 3 recordings are slightly out of sync, and why technically all battle themes are in 4/4 time signature. It was probably the solution that was the least trouble. Hopefully you didn't pull out too much hair trying to decipher the time signature! There's always multiple ways to solve a problem when making a game; sometimes you just pick one that's good enough.
Scrolled down to say basically the same thing, and well put. The reason this isn't pointed out more often is because the difference that audio processing causes is often so small that it's negligent and goes completely undetected.
@@chupathingy13exactly. Supposedly humans can't detect things faster than 13ms, and a 60FPS game means the frames are around 16-17ms.
But when it comes to audio, we absolutely do hear a stutter, Even if it's a lot smaller and just a little pop or crack. So usually they separate audio from frame processing.
Fun fact, if you've ever had a piece of software crash, and it keeps making the last noise it made, I think this is why. The audio processor is still going, but it's not getting instructions from the rest of the software on what sound/note to play next.
@@kantackistancan confirm from knowledge of nintendo console architecture, this is all exactly it. The GBA streams raw 8-bit PCM audio from the CPU most of the time, unless it's using gameboy emulation features. So all audio playback is at the mercy of the CPU's timing for game code.
It's possible, if not likely, that two recordings from console with different gameplay states wouldn't even sync up 100% over time. With little microseconds of lag every time a button press is registered and it has to branch to its process.
I’ll note that on older hardware this was even more suspect because frequently the audio chip and the cpu were not so separate. Super Metroid on the snes for example offloaded some processing to the audio chip, and thus doors can only open when sounds have finished. This actually made its way into the tool assisted speedrun for super metroid, as a significant consideration for optimisation. I assume the GBA had a clearer distinction between audio and game processing, but it’s totally possible that the audio chip runs slower the harder the cpu works.
It's also possible if not likely that recordings from 2 different GBAs would have slightly different playback speeds. The crystal oscillators that all signal timings derive from are intended for consumer devices, not scientific applications, and have some tolerance for deviation from nominal speed; I wouldn't be surprised if discrepancies of ±0.01% between units at identical temperatures are observable.
The way it's programmed to handle the heartbeats is fascinating to me, I know that several other games have sound engines on the GBA that use MIDI files in this way (hello everything at Game Freak), so it's unsurprising that they'd use them for this. But I didn't expect the *hard coded* use of quarter notes, that Shogo Sakai and/or a sound engineer assisting him would then have to convert into something the game could then use for its own system. The fact that this also makes the nightmare of Masked Man into 4/4 makes me wonder if Sakai *was* also the sound engineer at the same time he was composer (which would track, as a lot of older soundchip hardware for NES and SNES had the same thing going on) because this is too elegant to be handled by two individual people for how monstrous it is to our ears.
I don't know Mother 3's internals, but as a programmer... I think it's hardcoded and less of a limitation, as being able to effectively listen in for different note types isn't theoretically hard to do if you have access to a full MIDI track on actual hardware. It's more of a "build to scale" thing - Mother 3 has a lot of tracks, like over 200. Not all of them are used in battle, but many of them are or can be at certain times, so your beat tracking system needs to be robust to account for all of them. It was probably just a shortcut and the creation of a 0 volume MIDI track with exclusively quarter notes appended to the end of each bit of music data, that every song then needed to be engineered around after the fact. It's for a similar reason that all the instruments used by the Bash command are in C minor - though that also has both lore reasons and limitations on what kind of battle themes could be composed. Sakai is a wizard.
I think Strong One's variants are supposed to *sound* like 29/16 and 15/8 respectively, and I agree with the conclusion that we can treat them as "approximants" of these time signatures. Less because of the limitations of the GBA, but more because that was the artistic intent to be "just off" of them to make them harder to follow! After all, if we approached Dry Guys *not* as 3:2, musically that makes no sense, even if the game's actual data speaks otherwise. These slight tweaks are intentional because of Strong One's purpose - if Sakai wanted to make them 29/16 or 15/8 as opposed to...whatever this tempo map is, he could've, clearly. (I'm not in the man's head, but yeah lol) He wanted to mess with us a bit :)
...WAIT HUH LOL THE REFRESH RATE JUMPSCARE
Great overview, this was a joy to watch!!
EDIT: Thanks for the like :) Saw a typo I had to fix but UA-cam gets rid of it on edited comments, ah well
Someone in another comment said that the GBA Pokémon games use the same sound engine as MOTHER 3, which would explain why the heartbeat 'structure' is so similar.
I cannot *believe* the thing about the heartbeat tracks all being 4/4 quarter notes and hacking together anything else with tempo changes. What the hell.
I love how everytime you talk about something in music theory you then add that to your outro music because it makes me realize that without you explaining it I am STUMPED
My friend just shared this and I said to her, "this game taught me polyrhythms I swear".
Lmao gotta love the combo system
9:59 Imagine you're a composer writing for a videogame and suddenly you have to do weird time signature, speed and rythm calculations so that it's percieved as written above
71907/40120 is mathematically equivalent to the 4/4 beat changes. With 71907 beats in a bar, heartbeats 1,2,3 and 7 get 10,030 beats each, heartbeats 4 and 5 get 7,021 beats each, heartbeat 6 gets 5,355 beats, and heartbeat 8 (which gets divided into a triplet) gets 12,390 beats. It doesn't actually matter what the beat denomination is, but if you choose 71907/16384, the tempo (whole note) is the same BPM as the framerate in FPS. I don't know what the other math guy is thinking.
*71907
Okay, but it's not 79107/40120, it's 71907/40120. You added an extra 7200 beats there.
@@pbjandahighfiveit was a typo-
Taken from another comment:
So, interesting math note:
You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3.
You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain.
In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8.
The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song.
In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536.
In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576.
Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go.
(p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
oh i heard my name
I chuckled a bit when I heard “verified by the official earthbound wiki.” =P It’s a fan-run wiki, so it really has no authority to verify anything. They just curate information that’s believed or discovered to be true by superfans.
I mean, for stuff like this, that's about as close as you can get, especially if a source/proof is provided.
At the end of the day, any sort of "official" verification depends on trust of the supposed authority.
ARTSYOMNI!?
not even wikibound, a fandom one
Well...if you want to really break the term "verified" down like that, no authority on Earth can truly "verify" anything. I've found the Earthbound/Mother community to be very passionate about the series, so if any low-level discoveries were to have been faked or misreported, the community would pounce on it. I'm sure you could do your own investigation into the matter and see whether their findings line up with the facts.
Besides, not to bring politics into the discussion, but the US's CDC during the COVID pandemic made very bold "verified" claims that turned out to be false or made up on a whim, such as the 6-feet social distance rule being allegedly necessary and based on studies.
@ what I’m saying is that for information regarding a video game to be verified, it should come from a first hand source of some kind. For example, fans can speculate about the Zelda timeline, but if Nintendo, Aonuma, or a licensed publisher working directly with Nintendo offer up information about the Zelda timeline that coincides with that information, the information is now verified.
Sadness and Sorrow in that time signature at the end killed meee
What are you talking about it's just 4/4 /s
I was hoping someone else heard it too. I felt a little insane.
The "everything is in 4/4"joke was actually correct, cant belive it
5:52 Hey, it's me! I added those a long time ago. Quick note about the wiki, it's not official by Nintendo, and even then the wiki and its community have abandoned Fandom ages ago and are now an independent place, WikiBound.
Years ago I was playing with the game's memory addresses and found the address for the current BPM -- I have a video showcasing the value and what happens if you change it. That is where I got the BPM values I added to the wiki from. I do however remember that they never exactly matched real life; I think I vaguely remember checking that Zombeat is internally at plain old 120 BPM but when you actually hear the song the BPM is slightly different. Most likely because of the GBA's clock like you said, but also misc. stuff like limited decimal place calculation capabilities and lag frames.
I have not looked into how the game works very deeply, but I did gather some conclusions back then, and have made use of them to developed a since-abandoned project, PK Rhythm, which mimics Mother 3's combo system. In both Mother 3 and PK Rhythm, the system is pretty much what you describe: whoever made the song specifies BPM changes and then the game just checks if the player's hit is on-beat or not. Both are completely ignorant to the concept of time signatures, off-beats, etc.
This song is the true embodiment of how it feels to sightread for honor band auditions...
Cadence, as a former musician (I say former because I never get to gig or teach anymore), this is exactly the kind of content that me and so many other aspiring music professionals wanted to make and watch when we were younger. The fact that you are actively researching, making content, and experimenting with music so earnestly means that you are on your way to great success as a content creator and musician. Keep improving, keep learning, and keep writing. Great video!
Oh my God this is one of those times where you come across a video and feel like it was made specifically for you and you alone.
This was fascinating, thank you so much
Me as heck... I was the first person I am pretty sure to upload a 16-hit combo of this song LOL
Indie game dev here. Chances are the reason why it's not exact is because you have a variable amount of time that each frame is being processed before it gets to queueing up music. Let's say Frame 1 it starts immediately, then 30 frames later it's time to queue up the next bit of music. Before it gets to the music, it may be processing some other chunk of data such as player input that takes some milliseconds. So then the music gets queued slightly later. If the code is looking at the projected end of a string of notes, the next sections all get pushed out slightly each time this happens.
Ah.. I never even thought about changing tempo within the groove of the song. Holy cow that's crazy.
For those wondering where the 71907/40120 time signature came from:
In the actual sequence data, the tempo is not consistant at all.
The tempos are:
3/4 at 126 bpm
2/4 at 180 bpm
1/4 at 236 bpm
1/4 at 126 bpm
1/4 at 102 bpm
The total length of time for this sequence, is precisely 23969/7021 seconds (that's the simplified fraction).
It should be assumed that the intended tempo of Masked Man is 126 bpm.
Now, if something is playing at 126 quarter notes per minute, and each measure is 23969/7021 seconds, how many quarter notes is this?
At 126 per minute, each quarter note lasts precisely 60/126 seconds, which simplifies to 10/21 seconds.
Now if we divide 23969/7021 by 60/126, we get 3020094/421260 quarter notes.
This simplifies to 71907/10030 quarter notes.
Dividing 71907/10030 by 4 gives us 71907/40120.
Even the regular Strong One isn't _exactly_ 15/8. It's something like 14.88/8.
Taken from another comment:
So, interesting math note:
You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3.
You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain.
In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8.
The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song.
In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536.
In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576.
Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go.
(p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
This checks out. I use this formula and the result is the same:
(126/126*4 + 126/180*2 + 126/236 + 126/102) / 4 = 71907/40120
I'm not convinced that the interpretation of "it's 4/4, just speeding up and slowing down" is a useful one from a music theory perspective. Sure, that's how it's implemented in-game, but as you point out with Dry Guys and all the other tracks, you can really do any beat that way, even ones that are obviously intended to be in a totally different metre.
That said, the fact that the ratios make it a good distance from 29/16 is pretty convincing. But you also raise a good point at 11:15 in that it MIGHT have been meant to be 29/16, although the extent to which it's off makes me wonder if it was actually meant to be something different even from that. It does make me wonder what might be a good compromise between the somewhat-off 29/16 and the patently-ridiculous-but-technically-correct 71907/40120 I've seen floating around.
Taken from another comment:
So, interesting math note:
You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3.
You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain.
In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8.
The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song.
In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536.
In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576.
Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go.
(p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
Sounds to me like Strong One was written in 15/8 but as they were putting it into the game somebody messed up the tempos, and they thought "hey wait that kinda sounds cool, what if we use that for a harder enemy encounter?" The best creative decisions are often the ones found completely by accident.
Composer: Ho ho, this one's going to throw people off. They think it will be in 7/8, but it's actually in 15/16!
Intern: Actually sir, your calculations are a little off. It's actually a 43875x10^15/64 Time Signature
Composer: Ah it's okay. It's not like anyone will notice 😅
I love how the footage is you beating Lucas up, rather than playing as him 😂
i didn't play mother 3 but i think he gets bodied the first time he meets masked man so it felt appropriate
@CadenceHira Have played Mother 3 can confirm, then again [Masked Man] bodied me too (emotionally)
@@CadenceHira tbh the first time I met him he kicked my ass
I only had one guy left alive by the end
the fan translation has its own dedicated website. using reddit is just asking for malware, lmao.
Having ripped a fair amount of MIDI data from games, I can tell you that much of the time a time signature isn't even programmed and it's technically 4/4, even if the music itself is just in 3/4 all the way through. I also actually saw someone point this out about "Strong One (Masked Man)" somewhere, but people just got mad and insisted "well it's still effectively 29/16", which this video demonstrates is false.
Side note: that is not the "official EarthBound wiki" and just a fansite hosted on Fandom (blech). But the SiIvaGunner wiki is edited by people associated with the channel, so that one could possibly be considered official- for SiIvaGunner of course, not EarthBound.
The SiIva wiki is mostly handled by fans of the channel, though with how big the SiIvaGunner team is there ends up being a decent amount of contributors that also write stuff for it and are involved with managing the wiki.
i think the reason why the heartbeat notes are straight quarter notes with changing tempo is so that the combo sounds of the characters will flow seamlessly-ish into the next combo sound
As a percussionist, that drum notation at 8:15 made me physically sick.
A decent idea as to "why the in-game timing is slightly off from the expected ~59.7hz" is simply "its lagging a little". The GBA didn't have a dedicated sound OR graphics chip, so _everything_ went through the CPU. Graphics, Audio, Game Logic, you name it. Understandably, this is going to cause a little (edit: basically imperceptible to the human senses) slowdown - especially as the GBA gets older and hardware starts slowing due to age. Unless you find a video someone took of their GBA in 2006, and even then to a lesser extent, there's going to be inaccuracy.
What’s insane is that even though the timing in this song is so strange after enough listening the groove starts to make sense! Repetition legitimizes, repetition legitimizes
Repetition legitimizes
This helps really illustrate the function of tume signatures. Its so easy to just wonder "why?" as a person who only ever dabbled in music and who was good at learning by ear.
I mean, you show how most people are wrong, but you show how you can be so precise and represent something that feels like ear chaos succinctly on paper and for a machine with a purpose, which is cool
Shigesato Itoi: ok this is gonna be one of Nintendo's darkest characters so we need to make sure we give Music nerds anyuerisms.
"It's not that bad... it's actually worse"
Really fascinating to see how it all works behind the scenes!
Idk what you're talking about disappointed, this song being in 4/4 with wild tempo changes is such a great rug pull, and also much less common in the context of avant-garde music, and ALSO more nicely responsive to the affordances of computer-played music. Loved this video to bits! Thank you!
Seeing this is 4/4 feels the same as realizing power generation technology all just boil water in very weird ways
I was JUST listening to this song and counting along and while counting I knew something felt off about it allegedly being 29/16. I feel so validated.
I've experienced the effects of slowed down old VGM music cuz some sonic tracks have a funky tempo like 130.3 or whatever
i remember finding out about the battle songs being 4/4 with different tempos when i dumped the midis for the songs when i wanted to remix them, it was super interesting to hear fate/serious as it was programmed and i remember i talked about it in a random youtube commend like 7 years ago (will never ever find it again though).
that said i don't really think it's very useful to just say "in the end, it's 4/4 but with tempo changes!" (even if it's technically correct) especially considering that applies to literally every battle song in the game. the datamining of the music files and seeing how the tempo behaves surely gives some very interesting insight and it would have helped here, had the game not had that rhythm mechanic. like some other commenter said, shogo sakai clearly had a time signature in mind and i think creator intent should be what matters here. hell, for all we know it might have been his intention to make the song not fit in any steady tempo just to throw the player off even harder (after all it's considered to be notoriously harder to combo on than regular strong one).
if things don't sync up, especially in an old console, that can just be attributed to hardware inaccuracies, just as you said; after all you'll find that even NES games had trouble keeping a consistent tempo at all let alone a tidy whole number one. the fact the song seems to sync up really well at 29/16 at first tells me that 29/16 is a perfectly valid time signature to classify this song and i don't think it's worth diving into why it doesn't *perfectly* line up. lastly i have a feeling that the programmers simply felt like it was way easier for them to program the heartbeats to be in sync with the tempo and just told shogo "hey you'll have to play with the tempo lol sorry" who knows they probably wrote a script for him that generated tempos for him so he didn't have to manually calculate the ratios, but i don't think it was a hardware "limitation" but rather a "this is the easiest thing to code lol".
tl;dr for me this song is still 29/16 and if it doesn't line up exactly at that it can just be attributed to hardware and it's not really that important.
p.s. 5:14 fucking killed me you're hilarious
Taken from another comment:
So, interesting math note:
You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3.
You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain.
In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8.
The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song.
In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536.
In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576.
Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go.
(p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
Hahaha the siivagunner wiki was NOT what I was expecting when I clicked on this video randomly, got my sub now
This may be one of the most pointlessly pedantic music theory videos I've ever seen on UA-cam. I love it. Thank you for your commitment to quality.
Well, it could be said that time signatures in general tend to be approximate anyway. Classical music notation was never meant to exactly match the performance (nor the data from the game in this case), and neither using time measures (not talking about the 29:16 though, as it's still kinda manageable) nor precise BPM changes that are impossible to intuit feels satisfactory for a transcript, as like you said - it isn't helpful for performance - but I'd go further and say it isn't helpful for musical analysis, nor for any case in which the staff notation may be needed. I'd treat it similarly to other stuff that such notation doesn't usually notate (and even if so - never precisely), such as timbre or other things that appear only in a recording
That said I'm not a fan of notating it as 29:16 either as it still feels like there is a tempo change in the loop, but I'd probably just eliminate all but one of them by using approximations of the BPM ratio, possibly having a relevant symbol or annotation that the note length is a bit shorter/longer
Definitely true in a lot of cases, especially performance and stuff like you said. I think it's important in this context for two reasons: The tempos/time signatures are programmed in, and there was some level of deliberation in the exactness of timing that I think is worth analyzing, i.e. the feel if dotted eights:quarters is 4.29:3, made to deliberately feel like you're rushing. Also in terms of sync, any mashups or covers of this tune that intend to use the same beatmap need to know the temple cycle or it will desync.
Love this answer! Only thing is I would still argue 29/16 is sensible since the final eighth note in Strong One is effectively "cut in half"; it's the most straightforward musical representation of what Shogo Sakai did with the song's meter, IMO.
It is, cosmically hilarious that the people who figured out the time signature were the people who work on SiIvaGunner rips
the discrepancy in playback rate (0.58% vs 0.45%) is probably just due to lagframes
It's so cool how different representations in sheet music can give the same sounds! It's also neat how the music is *programmed* can end up very different from how one would write it for, say, actually making it playable for a live band, or comprehensible to laymen otherwise. Like, I doubt you'll have more people able to mentally conjure how rapid sudden tempo changes sounds compared to weird time signatures and odd note notation.
12:07 shit has me on the edge of my seat
You can approximately count it as "quarter, quarter, quarter, triplet, triplet, eighth, quarter, quintuplet quintuplet quintuplet" at 123.6 bpm, and it should loop nearly perfectly with the worst note being only 1/50 of a second out of place. In this interpretation, the time signature would be 211/120 (as 3/4 + 2/6 + 3/8 + 3/10), or ~7.033/4, which is only so complex because of the unresolved tuplets.
Taken from another comment:
So, interesting math note:
You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3.
You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain.
In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8.
The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song.
In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536.
In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576.
Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go.
(p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
Honestly this is such a great video. I know nothing about music stuff but I’m a huge Mother 3 fan and I’ve always thought something wasn’t quite right about that song.
Well I'll be damned! I was a little salty to not hear you talking about this in your "weird time signatures in Nintendo Games", but with this whole video dedicated to this specific song, I'm more than happy and, more importantly, completely wrong in my comment on your previous video! I'm sorry for the strife and thank you for making this video! I absolutely love mother 3 and hearing you walk through the technicalities of this song was a blast, please keep up your amazing work!
I cannot express in words how even though I’ve been into music with weird time signatures my whole life, I experienced severe psychic damage after hearing the drum part to this. I can’t even imagine it would sound good played by a real drummer
i listen to Haken and I've heard 23/16 out of them, this confused me
If you think of sheet music as a representation of ideas to guide musicians on how to play a piece (from the days before audio recording) then the actual music that comes out is the real source of truth for how it sounds, and that will vary every single time the piece is ever played. For example listen to any rock music with drums that wasn't recorded to a click and the tempo will fluctuate constantly throughout the song, but you wouldn't notate out the tempo change on every single bar (unless you have way too much free time) because it isn't useful to the reader trying to play along to the sheet music - you would probably just write "120bpm (played loosely)" or something like that
Computer music (particularly on older consoles) is in such a weird place because you have to define in strict mathematical form EXACTLY what is to be played and the computer will strictly interpret that using whatever method it has
The song would be notated for humans as 29/16, it would be notated for the gameboy engine in 4/4 with tempo changes, and in reality it is whatever the gameboy decides it is when it processes that information (which might technically vary on each play through, on each different gameboy, each time the temperature of the circuit board changes, etc), which again is different to what we would hear on a PC/Phone after ripping it from the gameboy
TL;DR - Reality is fake and we live in a simulation
Taken from another comment:
So, interesting math note:
You were able to notate Dry Guys using both a non 4/4 time signature and using tempo changes because both represent the same ratio: 2:3.
You can get this ratio by dividing out the common factors from each of the prime factorizations of the tempos until none remain.
In Dry Guys, dividing out the common factors of 96 and 144 leaves us with 2 and 3. And 2x3=6 -> 6/8.
The issue with Strong One and Strong One (Masked Man) is that their tempos don't share many common factors, leaving a lot of numbers to be multiplied together to get a ratio that can represent each song.
In Strong One, from 120, 172, 224, and 80, you can only divide out 2 twice. This leaves 2, 2, 2, 3, 5, 7, and 43. 2x2x2x3x5x7x43=36,120. Thus, the smallest time signature you can accurately represent Strong One with is 36,120/65,536.
In Strong One (Masked Man), from 126, 180, 236, and 102, you can only divide out 2. This leaves 2, 3, 3, 5, 7, 17, and 59. 2x3x3x5x7x17X59=631,890. Thus, the smallest time signature you can accurately represent Strong One (Masked Man) with is 631,890/1,048,576.
Both of these time signatures are very unfriendly to humans, so approximating with 15/8 and 29/16 or using tempos to represent these songs is probably the way to go.
(p.s. The slowdown you mentioned at the end of the video would not change these time signatures, since a uniform slowdown would not change the ratio.)
This is insane! And SiIvaGunner gets woven in as well lol
This entire video made me think of the “Nuclear power plants just boil water” post. Such a confusing song turned into nothing but 4/4.
Every time I hear someone talking about time signatures, at least as they relate to video games, I am brought back 15 years to a top 10 list from GameFAQs, top video game songs in weird time signatures. The number one place was attributed to Tor, the final boss theme of Iji (one of the greatest, most underappreciated indie games of all time), being in something like 27/16. I may not remember it exactly, but its something I have been haunted by these last 15 years.
Now I want to see 8-Bit Music Theory’s reaction to this lmao. Also Adam Neely. This is like landmark musical bullshittery
Me too, especially when 8-bit Music Theory did his video on odd time signatures, he only did Strong One.
I feel vindicated. There was a period of several days where I would spend 30 mins-1 hr listening to this piece at a time, trying to figure out its time signature in light of inconsistant claims, and the conclusion I reached by listening to the music was that it was in either 4/4 or 8/4. 4/4 seems correct to me - there is more emphasis on the fifth note of the tempo cycle than I would expect in 8/4 (beats 1 and 5 seem equal in this piece, where they wouldn't be in 8/4), but I can see the argument for 8/4, being that each measure now contains one full tempo cycle & it is no longer awkwardly split across every pair of measures.
Thank you for making this video. Hopefully this will set the record straight on how this piece works, in a musical sense.
(minor edit improving readability)
Feels so wierd to re-listen to Strong One without the 127383638353736383863662782628 tempo changes
Now, we need Stronger One. A song with an actual absurd time signature.
12:00 well, sound emulation will never quite be 100% fully accurate so if you want the real ratio then there's no other solution than plugging an AUX cable in the headphone jack of a GBA and capturing the audio yourself otherwise, you're going to have discrepancies. I have this problem all the time when trying to fit SNES music to the grid in a DAW, it just always drifts
If anyone has watched this video and is now super interested in Sakai-san's music, it's worth noting that a few days ago he released a new song, "Necos Calm Me," on streaming!
Also, that drum fill is NASTY.
As a computer science major, I'd say this may be a case of how time is calculated in the game's programming. To my understanding, some games in the past counted cpu clock cycles to time things right. If you go back far enough, you see games where the programmer had to literally calculate how long each instruction would take to execute, which is why many games in the 80s lacked background music during gameplay (think Galaga and pacman), because you can't predict what instructions specifically will execute, and thus can't time out music. Now, later on, counting clock cycles became a subprocess ran by the hardware itself, and all the programmer would have to do is check how many cycles have passed. However, there remains two problems:
1. The cpu clock on a Gameboy advance only has to be consistent enough to execute instructions. Beyond that, there's no need to make sure all manufactured cpu clocks run at precisely the same speed, which could explain why your recorded audio is different.
2. The point in the code at which the programmer checks the number of clock cycles may differ for the same reason as before: whatever actions the player takes may cause the check to occur earlier or later in the game loop.
Now, in modern computers, we have much more precise clocks, better methods of checking cycles, and can even synchronize with remote atomic clocks. But back in the day, that kind of effort just isn't worth it when the yield is a perfect time signature, especially when the bottleneck is fitting the music in the cartridge itself. So I think your last theory was most likely correct, and you can chalk the unsyncronized nature of your last recording up to either programmer or manufacturer error, not musician. Great video btw!
Edit: I almost forgot, many games synchronize the gameplay to the display refresh rate. Same principle, except instead of counting clock cycles, we're counting frames.
Id take it as 7/4 and leave it and never worry abt it ever again tbh. It sounds a lot cooler with that kind of meter anyways
I think the main reason the song in game doesn't line up 100% with the midi data is purely due to:
1. The GBA having inconsistent framerate count with audio. (This is stated in the video)
2. The game may either purposely or unintentionally slow down the song because it has to change the tempo frequently, but it must also keep the heartbeat on track to make sure the timing doesn't desync as the battle progresses.
It seems hard enough to keep track of the songs time signature in one loop, but in one battle it may loop several times. Therefore, there is likely something in place to ensure that the heartbeat doesn't loose track of the song itself.
it makes sense that the heartbeats would always be quarter notes. your heart speeds up and slows down - it doesn't play dotted notes. bpm changes are how our hearts work.
as a cs major, your first mistake is assuming that the gba runs at a 100% consistent framerate. Most computers will fluxuate a lot in their performance. And while the average is 59.90089 whatever, the gba needs to compensate for its lower bound of performance. In short, the music is likely truncating that 59.90089 acerage down to just 59, and playing the music at a tempo that will work for that fps (hence 0.58% slowdown)
That's a 14 minute video explaining how this extremely cursed sounding soundtrack is... ACTUALLY CURSED!!?!?!?!?!?!
This might just be the closest approximation of "cursed knowledge" I've ever come across.
Thank you for tainting me with this.
You know, i just thought of your other video on this game, around a day ago. And then THIS SHOWS UP?!? Love your content!
My knowledge of music theory was 0 before watching this, and i somehow feel like i know even less now than i did before...
I'm gonna be honest, I actually hear it in 7/4, or really a 2 measure grouping of 3/4 + 4/4. It could be an additive time signature but it'd be irritating to notate (and read) so I'd literally just give it the time signature (3/4 + 4/4) (in parens and everything) and alternate meters without putting another time signature down. It just has a hemiola-like effect in the 2nd bar. First measure is just 3 quarters. 2nd measure in the group is: dotted-8th, 16th tied to 8th, 8th tied to quarter, triplet 8ths. There's no rubato. There's no crazy time signature. It just feels weird because it's obscuring the beat in the 2nd measure of the groupings. There's nothing to hold onto until that last triplet. I even notated it the way we'd typically have done it in publishing to not *visually* obscure the beat if it were notated.
The timings of those rhythms are off with what is actually happening. It's a decent approximation but it very quickly desyncs, as I demonstrate in the vid. The actual tempos were datamined from the actual game files and the ratios are calculably different from the 29/16 (or 7/4) approximations.
You are correct that, at least in this case, it simply used the quarter note for the entire piece. In order to facilitate the metric changes it adjusted the tempo. We did this a lot in early notation software when it was difficult to change the note-value that received the beat (for playback or rudimentary mixing), and I suspect this is similar (or literally the same) to how MIDI is programmed. That being said I don't fully understand the nitty-gritty of MIDI programming, but I suspect it was due to the available resolution of the MIDI timing.
If you're looking at the game data files, I'll have to believe you on that. You don't actually show that (the MIDI file) in the video. You're not importing the MIDI into notation software are you? Just looking at how it notated rhythms it seems like you might have? Quantization settings can mess with that, and older MIDI and notation software were notorious for not working together well. If this is what you did I would strongly suggest you rethink. I worked on some film scores having to import MIDI and you frequently had *slightly off* meters like this, because the software quantization needed tweaking to get a best-fit. Heck I saw that all the time trying to import MIDI. I'm still unsure about the time changes you get from the MIDI. However, you truly never know what can happen with MIDI, as it can be performed live with a controller, and any notation-import is a translation. Sometimes a digital (exact) translation makes the least sense because the MIDI timing is, for notation software, comparatively continuous, not discrete. This is why you have to tweak that quantization. Traditional rhythmic/meter notation, which I believe this discussion is about, is symbolic and not continuous and if you took any performance and used a high level of granularity (e.g., milliseconds) in determining meters, rhythms, and tempos you'd get crazy stuff, not even counting expressive changes.
I just always try to look for the simplest explanation. This discussion really gives me deja vu of MIDI quantization issues. Just listening to the track itself, I'm not sure desyncing is actually relevant to the notation, but I think you're correct it's some kind of incompatibility between the timing of the MIDI and the game/engine frame rate. You might be interested in MIDI beat clock and timecode regarding possible explanations here. Beat clock (more specifically PPQN/MIDI resolution) might even explain why the tempo changes might be necessary in the programming if I'm understanding it correctly.
Perhaps I'm just used to interpreting clear-cut, rounded boundaries in meters (at least of non avant-garde stuff) and my ears are biased. I'm not a MIDI expert but I suspect that someone who is would be able to identify the incongruities.
I always thought that this was supposed to represent how the masked man is actually Lucas' brother. Seeing your own brother, even if you don't recognize him, turned into an emotionless cyborg is gonna throw you off.
This channel never disappoints me. Way to do an insane amount of homework on this one
I may not be the first
I am not the last
When I see Cadence hira
I click very fast
And I may be late
It might be fate
but I see new post
it feels like eating french Toast
Man I always wondered about this song when I was playing, it was so weirdly out of tune but somehow in-tune at the same time, so I never got the timing right
Great video! Love to see some Mother 3 videos and that little reference to Rhythm Paradise at the end
Leave it to Siivagunner to unravel the musical mysteries of an obscure GBA game that was never even released in America lmao
Halfway though the vid, I was starting to sniff sound card and CPU clock weirdness. I'm really glad you also thought of that and addressed it because I wasn't expecting it haha
Ok, this is what I understood from this piece.
I think the melody was thought of as two groups of three notes. In the first version, the first group is a triplet of quarter notes and the second group is three eighth notes. In the second version the two groups of three notes of the melody have been altered in tempo, leaving the fragments without melody unchanged. The first three notes can always be thought of as a triplet at 120 on the metronome. In fact, this makes it clearer that it is a very small change and is well conveyed by the small difference between 126 and 120. The second part at 126 "without melody" does not come in after the triplet has taken place regularly, but a a little earlier, giving the idea of two "overlapping pieces". After that, the third part at 102 can be seen as a very rushed version of the group of three eighth notes, in fact I would write it as 3 eighth notes at metronome 204, rather than as a triplet.
Edit: my theory is that the creator had that specific music in his head or had recorded It previously and just found some tricky way to aproxímate It with the software that they were using and make It sound similar to what they wanted and It somewhat worked
Hey i just wanted to comment that your videos are super super interesting. I subscribed from another video but since then you published a few more and didnt disappoint at all!
"it's always 4/4 with tempo changes" is funny
Okay that Rythm Hell part at the end sent me flying, all in all a great one, your video :D
5:07 I literally GAGGED when i seen that
Adding the battle tempo to the music makes it feel like you're walking then running then tripping over your foot
it's disappointing that this video blew up in less than a week but you only have 33.5k subs, this is great video essay content!
This alone makes the 16-hit combo achievement on retroachievements seem impossible
The outro of this video is HILARIOUS LMAOOO
excellent stuff!! Quickly becoming one of my favorite YT channels!
"The difficulty of the battle, apart from damage and health pools, is also dependent on how cracked you are at rhythm games." *_DICKO MODE JUMPSCARE_*
good catch lol
did you ever consider irrational time signatures? it sounds very much like one to me. it would be exactly what you wrote at 3:14, except that the first six beats would consist of a bar of 6/4 (or 24/16), and the last three notes you wrote as "3:5 sixteenths" would actually be placed in a separate measure of 3/10 and consist of 3 eighth-notes. talking in the original meter, those three notes would be played at the speed of eighth-note quintuplets (just with only three of them instead of the expected five), but it would be cleanest to write them in the way i mentioned. if you really wanted to put it all in the same measure it would have to be 36/20
people saying "all music is 4/4 if you don't count it like a nerd" as if changing the tempo multiple times per bar isn't some advanced music nerd shit lmao
it's not music nerd shit it's programmer nerd shit
this is my first video ive seen from u and i love it lol, i like music theory and love older games. the "void my applecare" line was funny