Posted by simonw 2 days ago
In this song, which is also chapter four of the movie Interstella 5000 movie (spoilers from here!), the knocked-out singers are scanned, parameterized, brainwashed, uploaded into The Matrix, and then used in the following songs of the movie-album to robotically mass produce music.
It makes perfect sense that the BPM is 123.45 because that’s exactly the sort of thing you get when a manager (who’s shown at the end!) just enters some numbers on the keyboard into the bpm field. They don’t keysmash the numpad; they just hit 123456789 until the field is full!
So not only does the song itself convey what some boss thinks is music, robotically beating at 123.45 bpm, but it is itself about being endlessly-rotating brainwashed-boring cogs in a pop music production industrial machine. I’m pretty sure the movie scene cuts and animations are timed specifically to the beats of the song, but knowing that they’re timed to a machine-specific bpm that a human would never select at random with a metronome?
Absolute genius.
I had no idea. Thanks for posting this.
EDIT: At 123.4567bpm, I think the track has precisely 0.2345 seconds of silence before the first 'beat' of the song and actually has 456 beats total, which is either numerological nonsense or pure genius by Daft Punk. Math elsethread :)
This seems like quite an assumption. Why wouldn't they keysmash? Or make up a fake number? And why bother to add a decimal point? What is meant by "robotically beating at 123.45 bpm"? Any fixed tempo beats robotically.
Your theory could be correct but it feels like connecting too many dots to me. 123.45 is a bizarre (and kind of human in that way) tempo that strikes me as more of a cheeky easter egg than a deeper connection to themes of corporate mass-produced roboticism (if they even did intend that as the exact tempo).
Couldn’t resist the dad joke. In any case I was enjoying you enjoying DP
Perhaps they wouldn't need to? iirc Modern MPC you can enter 12345 on the BPM touch entry field and it will fill that in as 123.45
While a robot can keep beat at 123, most humans can’t keep 123.45. Art doesn’t have to make logical sense.
Isn’t it also true that while a robot can keep beat at 123.45, most humans can’t keep 123?
Apart from training, there isn’t anything that links human biology and psychology with the length of a second, is there?
Specifically about keeping tempo, human drummers don't really. They will move around a central tempo, slowing in verses and increasing tempo in choruses and as the song progresses. If you're hearing a fixed tempo in a song, it's because it was recorded with a click track in the drummer's ear. Super common these days because popular tastes for recorded music currently skew towards perfection.
I wholeheartedly agree. This thread would have been a lot less fun to create if I’d had to apply rigorous methodology and proper hypothesis evaluation practices. I’m really glad someone else appreciated that too :D
The impression I got as a Daft Punk fan in the 90s was that the movie was commissioned alongside the production of the album and not an afterthought.
The album was released after a couple of singles (iirc) but that’s very typical for artists to do. So it would make sense for the movie to also be released after the singles, even though it was already (mostly) completed.
Edit: seems my memory is largely correct. The movie was always a planned part of the album.
https://en.wikipedia.org/wiki/Interstella_5555:_The_5tory_of...
Production of the movie started right at the end of the lengthy recording sessions (reminder that Discovery started being a thing late into the Homework recording sessions, with Short Circuit & One More Time being early productions) but it's rather uncertain how this process started. The one thing that I'm 100% sure of, is that the entire production was funded by DP themselves, and they are probably the only owners of the golden master of the movie. (Hence their media team's fault about the absolutely crappy 4K remaster and noise around it, that led to that awful Epic collaboration).
Singles were released at rather odd intervals for Discovery in particular (according to Bangalter, all songs were conceived as potential singles) and music videos were released after the singles.
EDIT : To be honest, it's a bit tricky to be 100% sure about the actual process, with most of the claims being unsourced random internet shenanigans, and the tendency of some people (hi Pedro) to rewrite history whenever they wish.
There is no way the anime and the movie wasn’t part of the original albums concept and neither the part you quoted nor the link I shared disagrees with that point.
I agree the album is an album of singles. But that was never in dispute.
I also deliberately avoided discussing the themes of the songs and plot of the movie but musicians are frequently reinventing meanings to songs. And if the history of operatic movies has anything to go by, it’s that musicians seldom have a coherent plot for these things (regardless of whether you’re Philip Glass or The Who) so it’s usually better to just enjoy the randomness of these movies for what it is. But that doesn’t mean that the anime wasn’t still part of the original album concept.
The earliest copies of music videos being sent to stations are dated from around February 2001.
Discovery was released in March 2001.
The first single bundled with a music video was the fourth one and was released in October 2001.
The earliest broadcast of the four initial videos seems to be August 2001.
So no, music videos and single releases were not really bundled together.
Those dates could more easily be explained as delays in the animation process.
I’m feeling unwell though, so apologies in advance if I have completely misread and misunderstood your post.
It's more then likely the backstory he outlined, which is I believe a minor subplot non-essential to the main story of the movie - has been added like this precisely because that was the theme of the song.
Because this was actually made by humans, they frequently talk with each other when making art in collaboration
Discovery has been explained many times by DP to be about childhood, not having any specific "theme" besides mixing disco and rock. Hence the name "disco very" and the "pun" in Veridis Quo. (which also happens to be a major sequence in the movie. Although DP never cared to enter the details of that particular composition, most likely memory hole'd by the protagonists.)
So no, this is definitely not the theme of the song. There are several years between the actual songwriting and the release of the movie. Heck, if you actually see the movie, the ending sequence kinda explains that this is "one" of many interpretations of the record...
Taking a look at past interviews, it is more likely that 5555 is about what surrounded the actual release of Discovery (hugely anticipated sequel to a magnum opus that was wildly different from expectations) rather than an idea that was here from the start; see also Human After All for a continuation on this theme.
I can't believe I've only just learned about this
Anyway that album, Discovery, is full of funny bits. Track #11 Veridis Quo sounds like "very disco". Turn those two words around, and you got the album's title.
Haha if only
Well the tempo is steady by human standards, but latency and jitter on timing signals are recurring issues in electronic music. Some devices put out very steady timing but don't like being slaved to another device, bugs can creep in at loop points or pattern switching (even on Roland's latest flagship drum machine, which costs most of $3000), things can get messy if there is too much note/controller data and so on.
But it's not about audio jitter.
Anyway, like I said, too much speculation in this thread.
Yes, MIDI jitter can be compounded on the receiving end - but having a very tightly bound MIDI clock to the audio clock can negate a lot of those issues upstream in the first place, and that is precisely why you get a good audio interface that does this anyway.
(Disclaimer: have worked in pro audio product development for decades, have written drivers for exactly this use-case, and I have personally been in the trenches to fight the myths about Audio and MIDI jitter as a developer for a long time now..)
Reasons are:
- Someone here commented that Reaper gives 123.47 BPM
- I implemented my own BPM finder back in the days, which is quite accurate with electronic music that doesn't change tempo, dug it out, and got a result of 123.48 BPM
- I looked for files for rhythm games, and I found this https://osu.ri.mk/beatmapset/1495670 with a listed BPM of 123.48
If you want precise BPM, I suggest looking at rhythm games (DDR/ITG stepcharts, OSU beatmaps, etc...). People playing these games really want tight timing, in the order of 10 ms or less, it means that a difference of 0.01 BPM matters. For top players, a difference of 0.03 BPM would be completely off after a couple of minutes.
There's a better visualization of the track here:
Also, C418 put a creeper face in Minecraft's soundtrack.
If you're up for it, trade a music rec?
Try:
Scorpion Mother - Thief https://www.youtube.com/watch?v=5A3113EQvLg
Certain Indian music and metal seems to scratch a similar itch for me. And of course orchestra and drum n bass.
60 * 445 / 216.276 = 123.453365145
60 * 445 / 216.282 = 123.449940356
60 * 445 / 216.276 = 123.453
60 * 445 / 216.282 = 123.450
So if the correct pair of values there ends up being 445 / 216.27000197, then it'll be:
60 * 445 / 216.27000197 = 123.456789
Or, since one of those programs had four decimals:
60 * 445 / 216.27015788 = 123.4567
Or, if it's 444/446 rather than 445:
60 * 444 / 215.78415752 = 123.4567
60 * 446 / 216.75615823 = 123.4567
But I see that they cut the "whooshing intro" at the front, which I imagine is part of the beat — they're in the hands of the machine now, after all! — so if we retroactively construct 123.4567 bpm into the silence (which, they estimate, is 5.58s):
5.58s * (123.4567bpm / 60s) = 11.4814731 beats
Assuming that the half a beat of slop silence there has to do with format / process limitations with CD track-seeking rather than specific artistic intent, we get:
+11 intervals @ 123.4567 bpm = 5.346s
Which, when added to the original calculation, shows:
60 * (445 + 11) / (3:41.85 - (0.5.58s - 0:5.346s)) = 123.4567 bpm
And so we end up with a duration of 221.616 seconds between the calculated 'first' beat, a third of a second into the song, and the measured 'last' beat from the post:
60 * 456 / 221.616 = 123.4567 bpm
Or if we use the rounded 123.45 form:
60 * 456 / 221.628 = 123.45 bpm
And while that 22+1.628 is-that-a-golden-ratio duration is interesting and all, the most important part here is that, with 123.4567bpm, I think it's got precisely 0.2345 seconds of silence before the first 'beat' of the song (the math checks out^^ to three digits compared against the first 'musical beat' at 5.58s!), and so I think there's actually 456 beats in the robotic 123.45 song!
:D
^^ the math, because who doesn't love a parenthetical with a footnote in a red-string diagram (cackles maniacally)
5.58s - (60 * 11/123.4567) = 0.2339961 ~= 0.234
5.58057179s = 0.23456789 + (60 * 11/123.4567)
I'm not sure but I think this is also the smallest time resolution.
Then each frame is composed of samples, but they seem to be counted in groups of 1/75th os a second anyway.
Besides the sample period, the total number of samples matter for frequency resolution (aka BPM precision).
44100 Hz sampling frequency (22.675737 us period) for 216.276 s is 9537772 samples (rounding to nearest integer). This gives frequency samples with a bandsize of 0.0046237213 Hz which is 0.27742328 BPM.
Any claim of a BPM more precise than about 0.3 BPM is "creative interpretation".
And this is a minimum precision. Peaks in real-world spectra have width which further reduces the precision of their location.
Edit to add:
This takes my flac rip of the CD and simply uses the full song waveform. This artificially increases frequency precision by a little compared to taking only the time span where beats are occurring.
It is correct to say "0.0046237213 Hz which is 0.27742328 BPM". My mistake was to interpret 0.27742328 BPM as the limit of frequency resolution in units of BPM. Rather, any BPM measured must be an exact multiple of 0.27742328 BPM.
Thanks for pointing out my mistake!
> (regardless of track length, which already smells suspect)
Frequency resolution being dependent on the number of samples is a very well known property of basic sampling theory and signal analysis.
In fact, one can interpolate the frequency spectrum by zero-padding the time samples. This increases the resolution in an artificial way because it is after all an interpolation. However, a longer song has more natural frequency resolution than a shorter song.
Note, this frequency resolution is not related to fidelity which is some messy human related thing that is over a sliding window of shorter duration that I don't pretend to understand.
BTW, the opposite is also possible. You can zero-pad the spectrum as a means of resampling (interpolating) the time domain. This is slower but more spectrally correct than say time-domain linear or cubic interpolation.
These techniques require an FFT and so are somewhat expensive to apply to long signals like an entire song, as I did for the plot. Daft Punk's HBFS takes about 8 seconds on one CPU core with Numpy's FFT.
> > So we've never actually made music with computers! [laughs] Neither Homework nor Discovery nor even Human After All were made with computers.
> Was he contradicting himself from 12 years before? Or did he forget? Or maybe it's a terminology thing?
The thing is—and this is coming from someone who has been making electronic dance music daily for over 35 years and counting—when Bangalter spoke earlier in their career about a PC (likely an Atari ST or Falcon) it was being used as a MIDI / SMPTE timepiece and master sequencer, nothing more. Later when he speaks about never making music with a computer, the context of the discussion has changed, as by that time computers were becoming more accomplished at DSP. The comment he is making is that they didn't use computers for audio domain tasks, like Pro Tools, Digital Audio Workstation type action.
That said, computers were still deeply embedded in their workflow just not in the way most modern producers would recognize. Even the SSL 9000 J console at the heart of their studio relied on an onboard computer system for total recall, automation, and channel configuration. The distinction Bangalter draws is really about where the actual audio lived: in 12-bit sampler memory, on tape and through analog audio circuits, not as samples and waveforms being crunched inside a CPU. The computer was a conductor, not a performer.
I was working in studios around Europe in the late '90s and if you said "Logic" in a studio context, you were certainly talking about Emagic Logic, and "PC" didn't mean a Windows box. In that era, particularly in France, "PC" was often used colloquially to mean any Atari ST or Falcon, which had been the backbone of European electronic music production for a decade. Given Daft Punk's roots in the French house scene and the timing of Homework's production (1996-97), there's a strong chance they were running Emagic Logic on Atari hardware, because at the time, the ports of this program to other platforms were garbage and were not to be trusted.
The lineage of the software is an entire saga unto itself. What became Apple Logic started life as C-Lab Notator on the Atari ST in the late '80s which dominated Euro electronic music. In late 1992, after a dispute with C-Lab's owners, the core developers, one of whom was Apple's own Gerhard Lengeling, walked out and founded Emagic. They rewrote everything from scratch as Notator Logic, which eventually dropped the Notator prefix and just became Logic.
Around '02, Apple came knocking and swallowed the whole operation. They immediately killed the Windows version, and dropped the Emagic branding entirely with Logic Pro 7. Like I said, Gerhard Lengeling is still at Apple, now their 'Senior Director of Software Engineering for Musical Applications' according to his LinkedIn.
Do not assume that because a particular gear prints out a particular BPM value that the actual BPM is that. Plenty of midi gear is inaccurate.
