@JM Coulon to be honest my area of expertise is composing using synths and sample libraries so for me it would be ideal to be able to have that level of control and precise instruction with a session musician if I needed it, but in practice people aren't really able to execute that, and it really isn't fun for them to try
a decent tactic that does sometimes work is to program a loop that has the microrythm, dynamics, and feel you want, and then say "do your best to match the feel of this" and give it to them in the headphones instead of a click.
Everything can be notated, but not always in a way that is purposeful for a musician to read. If your Dilla-groove is looking like a Ferneyhough piece something has gone wrong.
Don't wanna jinx it, but I'm so thankful that traditional copyright laws don't include drum melodies or grooves, so we can analyze drum legends to our heart's content :)
@@rzufig961 oh right that explains it. i was also like whoa, what just happened check out his screencaps (e.g. 8:55), turns out even his computer is in German.
Fun fact: dogs apparently have almost perfect sense on rhythm. Saw it on QI episode. They trained a dog like 118bpm means going for a walk, 120bpm means food. Now if I could only trained my dog to bark when the drummer strats in the wrong tempo! Animal metronome.
I hope you do a follow up on how this affects the "chemistry" between a drummer and a bass player. As a bass player I've played with a lot of drummers and sometimes it is easy to play together and sometimes it is really hard. Over time I've started to feel that this is mostly an issue of having or lacking a shared understanding of microtime. After all, if a beat is slightly behind the quantized position, this could mean two things: tempo modulation or microtime and both of these would imply a different time at which the next beat should occur. The same holds in reverse as well - I might play a note just slightly behind a bass drum hit as a microtime choice and the drummer might try to adjust for me dragging. Another possible follow up would be something on how to use quantized and unquantized grooves as a means to generate contrast.
Agreed. When a drummer/bass player finds their counterpart that they fit with, you just know. Can't put your finger on it, you just know the two of you got it.
Have a read into musical entrainment, there’s some academics doing pretty fascinating research on isochronous rhythms & dynamic systems. Some use the social/metaphorical application, beyond the standard “physics” parameters, such as how well the players know eachother etc. Always excited to see people exploring these changes in phase relationship
I believe the micro timing perception is likely going to be relative to the tempo in those tests. By increasing (or decreasing) the tempo, a fixed shift of the 5th note will represent a larger (or smaller) portion of the beat interval. This would be an interesting controlled variable to test.
Wow, that's a quite important control measure not thought of in this experiment. I guess the frequence/ pitch of the note also plays a role. Maybe a high-pitched "beep" that most metronoms give you will cause a better distinction - and headaches in all participants :D
I get unreasonably excited every time Shawn uploads a video. Man, you do such a great job breaking these concepts done in such a way that it is a joy to learn.
Bro SAME! I think it's probably because you've been grinding specific timings at 60Hz for OoT speedrunning, because I've been grinding my Super Smash Bros Melee (you'd enjoy the process of getting good at melee, I think) timings, and I was thinking, "oh, that felt about a frame early" and it was right in that neighborhood. So fucking sick to realize that your brain is capable of processing timing windows that quickly. Anyway, you're dope, keep up the good work my man.
As a self taught guitarist (15 years) and newbie bass player (4 months) I really expected to do terrible, but my scores were consistently 8 and 9 until 10ms. 10 and 5 were rough... this is all as fascinating as people's pitch perceptions. I love studies about this stuff, great video!
Excellent stuff! I reckon Shawn has already read the famous Finnish study from 2015 analyzing Jeff Porcaro's timing: "Fluctuations of Hi-Hat Timing and Dynamics in a Virtuoso Drum Track of a Popular Music Recording".
True, that would be interesting. The hitwindow on higher difficulties can get pretty low. I don't know about other games but in osu! OD10 is +-19.5ms, so pretty much at the threshold that can be consistently perceived, according to Shawn's data here. Although you'd have to survey some really high level rhythm gamers for them to compete with drummers with 10+ years experience ^^
In Rock Band, there were some very fast notes that I couldn't hit until I turned the calibration to the nearest 5-10 milliseconds. Anything outside that threshold and the game wouldn't register it.
I'm a student musician that plays violin, viola, drumset, and marimba. Been playing for around 7 years. I got all correct up until 30 ms Got 1 wrong on 30 ms Got 5 wrong on 20 ms Got 6 wrong on 10 ms And then 4 wrong on 5 ms My overall score was 54/70, although luck was on my side for the last one :p
Percussion, haven't played in a few years (apartments, man), played for about 13 years basically consistently until I was ~21, always in a student context. All correct through 50, 2 wrong on 30, 8 wrong on 20, 9 on 10, and 4 on 5.
i've pretty much only ever used DAWs, but played a bit of keyboard in the past. i got all correct up to 3 ms 1 wrong on 30 ms 2 wrong on 20 ms 3 wrong on 10 ms but only 4 right on 5 ms my total score was 58/70 maybe listening to lots of speedcore (~ 300+ BPM) has something to do with it?
I have no musical experience apart from attempting to play the first "Smoke on the water" riff and singing in the shower. Got 57/70, made one mistake in 30ms and 20ms, half wrong in 10ms and 6 wrong in 5ms. I did re-listen and check it for 2-3 times though, not sure if it's considered cheating.
You should do a similar experiment with rhythm gamers. I was in school band class for 6 years, and have been playing rhythm-based games (DDR, Beatmania IIDX) for another 20. Quantized beats are literally the objective. I got 100% all the way through 30ms just like the professional drummers, as did a number of other rhythm game fans. I think you're definitely on to something with that range of perception being trainable.
Hey Shawn, I felt like the comparison of the "flat and lifeless" sound of the quantized version vs the original version of Steve Jordan's recording was really unfair to the sound of a quantized beat. Wouldn't a large part of that be the negative effects of attempting to quantize raw audio? I feel like in order to have a significantly more fair comparison we should have either two raw audio recordings where one is played with the 20ms delayed ghost notes and the other closer to being in the pocket/quantized; or perhaps using 2 digital drum tracks where we can control with much more accuracy exactly where each hit is placed. I loved the rest of the video and have been a fan for a long time! Keep it up!
I agree... not trying make a case against quantized beats (though I understand if people interpreted it that way). that segment was simply meant to be "check out this example of microrhythm!" anyway, I actually did the MIDI comparison at the time of the original experiment (didn't include in vid only because of time constraints). here are some examples you can check out: drive.google.com/drive/folders/1h7eQcaDVxBvRba2Z2--TlVjTB8Mo9vQO?usp=sharing
Dude, props to your data interpretation. Most people who don't have a background in experimental sciences don't know how to properly interpret and - more importantly - criticise their data.
This would be interesting to have in a mobile app. Then you could see how your time awareness improves the more you work on it. I've actually been working on a mobile metronome app, which I could modify to do this.
I'm very happy with my results! 9/10 - 200ms 10/10 - 100ms 9/10 - 50ms 10/10 - 30ms 5/10 - 20ms 6/10 - 10ms 5/10 - 5ms After the 30ms I can't spot the diference. Can't tell I really heard (or even feel) if the beat was late or early. I'm 31 now, I had drums class when I was 14 and again with 16. Them I start playing acoustic guitar, had classes for 1 year and stoped. I still play, but only for fun and with no consistent what so ever. What I do is watch your channel and other channels related to music stuff and listen to music on a daily basis. I don't have a good ear for pitches, but I have a good ear for volumes, don't know if this help to understand my results. I really like this video and your channel! (I don't speak english but I tryed my best to type this)
I have very similar results and story... I am curious - do you drum with your hands on your legs, or desk, or anything a lot? I find it very difficult to keep still, and drum with my hands a lot like this and have done since I last played drums about 20 years ago. I wonder if this has helped my score?
Your research made me think about two interesting ideas that might explain the drop off around 20ms besides just the limit of human perception. 1) You should try to group your dataset by early/late samples. I have a hunch that early samples will be a lot harder to detect than late ones. I expect this from your choice of sample. While a very short sound, a snare hit is not a perfectly "point like" sound. It smears in time for at least 5ms. I think that even if you play it early (-20ms), since there is still noise at the moment we expect (0ms), we still perceive it as on time. 2) At the speed of sound in air ar 20C, 1ms is 34cm. I'm sure you know how people who hear themselves talking with a delay will have a tendency to slow down. What if this is what's happening in that last performance? You mentioned how the delay was more pronounced on the kick, it also happens to be the furthest away from your hear. This is a very subtle effect, and its generally known that under 5ms of latency in recording is not perceptible. But what if it is, only subconsciously? In this case, we're talking 2-3ms at best... what if we include nerve response time? Assuming that pain/reflexive reponses are the same speed as conscious programmed triggers (like when you try to follow a beat and you consciously know when you should move your arms to hit the next note at the right time), well, those are at about 15ms... together thats... awfully close to the 20ms delay we're seeing here. I feel like there's a lot of interesting questions to answer around this topic. More research is needed.
THIS is what happens when you hear a drummer play a couple measures and realize he or she is a good drummer. Like you said "even if we can't consciously percieve them, we can feel that they are there". Sometimes can't tell exactly what makes a good drummer (or any musician), but I think microrhythm is an important spark added to it. Thanks for this! and please, do not erase the test, I would like to make a video about it, giving you the credit, of course!
This is great that you bring this up. We producers usually play around with pushing some hits forward or backwards by samples. Basically very low ms. I had a test in production school where I had to listen to a test like this. We also analyzed songs that use this technique to make a song feel fast and jittery or dragged and relaxed. I had a kinda easy way to feel that 10ms faster note because it felt stressed to me. Deliberate dragging and rushing is extremely important to make a song have the right kind of feel like you showed. VERY overlooked sadly.. Max Martin is a MASTER production technique. Even with midi drums.
Drumming ABSOLUTELY changed how I "felt" music as I played it. I'm totally unsurprised we're well represented in this niche music skillset.... and was surprised by how confidently I could see myself taking this test... until you brought up drummers 🤣
I love everything about this video 🤙🏻 Everything is well explained and I like how you even went into the ms necessary to identify a difference. Microrhythms are the key to studying the center of mass of rhythms and seeing what feelings are produced when those small shifts in timing happen. There's all sorts of fun stuff there. Great work!
It’s kinda of a big relief to hear that I’m not the only one consistently rushing or lagging by 10-20 ms. I’ve often manually quantized the feeling out of my jams. 😂
This just came up in my feed and I remember taking the test when it came out. My thoughts this time is if hearing these different flams is teachable, the group to test would be snare players from the top drum corps and marching bands. One thing that dci produced during the golden years was methods for educating students. Certainly there are instructors out there with trucks for eliminating the flams between players.
Played in a drumline in middle school for 3 years. Results: 9/10 for 30 ms 10/10 for 20 ms 5/10 for 10 ms 5/10 for 5 ms Quite happy with the results. Can I microrhythm? Apparently yes!
I’m not a musician, but you made it easy to understand and appreciate the topic. (Kudos on presenting the caveats of your research too!) The quantized vs original comparison was awesome!!! Thank you!
I actually noticed that i was way more confident in my ability to tell when the note was early, even at 20ms, but at that point it was more of a feeling based guess of if it was late or in time, that seems like an interesting thing to note
I was trying to put a finger on this exact thing a few weeks ago when I was trying to explain to a friend why I love Anderson Paak’s drumming so much, it’s not flashy but his feel is so good.
Shawn, I've been following you since watching your videos with Adam Neely, and later you inspired me to learn Japanese (still in that journey)! Thank you so much! One thing I noticed about the microrythm test is that I changed HOW I was listening throughout the test. At first, I was focusing my attention on the attack of each beat to determine if the 5th hit was early/on time/late. However, somewhere around the 30-20ms mark, I realized I was better off listening to the entire sound envelope, paying close attention to the decay of each hit. That really improved my score: although I scored a 5/10 on the 30ms test, I scored an 8/10 on the 20ms! Something to think about! Keep up the great work,.
In the beginning of the video I was really suprised that most people could feel a difference of 20ms. But then again, the way we tell the direction a sound is coming from in part has to do with the different times at which our left and right ears recieve that sound. So there are actually a few aspects of our perception of sound that occur at really small time scales.
Not saying the topic of this video and the thing about direction are related. But they're both messing with me because I've never thought of perception as being that quick I guess
Extremely interesting experiment. 200ms, 100ms and 50ms maxed out, then it went downhill from there. Interesting fact - at 30 ms, I made 3 mistakes, all of which were assuming that something is in time when it wasn't. It took me an additional tier to start seeing ghosts. 10ms and 5ms are just pure random as expected. ps. primary music school (6 years) graduate on the clarinet. Nowadays musically active as a hobby.
New sub here, great video! Just worked through the test- 100% score until the 10/5ms rounds whereby my brain apparently switched off and I scored 1 and 4 points respectively.
Listening 1-4: 10 of 10 Listening 5: 7 of 10 Listening 6-7: 5 of 10 Guitar is my main instrument. Before the test, I thought that finding the early ones might be easier, because both playing behind or on the beat feels "good," while early beats feel uncomfortable to me. But actually I did all kinds of mistakes :D
One weird analysis I draw from this is the use of drummers even in the age of DAWs. Don't get me wrong. I started out with drums and am now a keyboardist and after watching tonnes of your and Adam's videos, have also increased my musicality and knowledge of theory and other things. The point is, like you said in the end that Jordan sir didn't consciously think before adding those milliseconds worth of delay, he just did it naturally, unconsciously. And that's what can happen naturally only. You can't program such a natural feel into DAWs. So no matter how better DAWs might get, Sungazer still needs you man! Just kidding, hope you get the point. Big fan sir!
Thanks for this contribution Shawn. Is't beyond interest and amusement, it is showing a human skill of adapting (like i call it) the universal swing. My abilities (after 40 years of drumming) end beyond 30, maybe 20 ms...... which explains my BIIIIIG tolerance for timing fluctuations. The whole topic explains roughly why my fellow musicians over the years - well-known and unknown - differ so much in the acceptance of timing variabilities and the evaluation of them, together with the judgment of musicality in relation to timing.......
On the production side, with regards to creating reverb, the old rule of thumb is that up to 20ms of time delta is perceived as reverb, after 20ms it is an echo. I also find it interesting that 20ms is roughly the maximum of input/output/roundtrip lag that any musician, professional or not, may find acceptable when playing their instrument - which is mostly of course a problem in the digital world only. And that's why we all will eventually use ASIO and external interfaces. I've also found that the more experienced I get, the more sensitive I am to being bothered by any such lag. When I was younger, I had no problem creating music with a keyboard/output roundtrip that definitely had closer to 30ms of lag. Now it is out of the question, even 10ms is only barely tolerable, momentarily, before I want to throw the PC out of the window. Also, for the results of your study, I think the actual distinction is people who routinely perform against a metronome and people who don't, rather than professional or non-professional musicians per se. But, of course, a professional is much more likely to have spent countless, hundreds and thousands of hours perfecting their ability to play against a click, otherwise they're probably not getting hired for anything. Anyway, cool vid!
That's so crazy I've been thinking about this topic all week because I notice even amongst world class drummers some seem to have better time accuracy than others
Did the test and the results pretty much followed the same patterns as the other professionals with 15 years plus experience. Really interesting. There must be a point where it becomes almost impossible to perceive the difference, as you suggested in this video. Nice work, always enjoy your videos.
I read a study for my college course on microtiming and it’s impact on Groove and against everything what many music TH-camrs say the study showed that perfect quantized recordings received a very high groove rating sometimes even higher than the original recordings. Although I should say that the „grooviest“ versions of the recordings did always have some microtiming.
The fact that I got any correct @ or below 50 ms is pretty crazy. One issue is the short time between examples. I was still clicking the answer when the next one started.
Please note that choosing "in time" for all answers for 20ms, 10ms, and 5ms (which I did because I couldn't hear any error) gives you a score of 50%, 40%, and 40% on each of those, _not_ the 33% expected from a randomly chosen result, so the survey results should possibly be analyzed/interpreted differently. Also, as someone who scored 70% on 30ms and as noted couldn't hear 20ms of error, I *also* cannot hear any difference between the quantized and unquantized Steve Jordan performance, not even as a subconcious "just sounds better" sort of thing. (Ditto the Rick Beato video where he also quantized a drum performance.) I am curious to what degree this is true for others, and how that breaks down between professional musicians and non-. And if it is different between them, what this says about the process of music creation when the creators have aesthetic opinions the listeners do not.
Very cool video and had a lot of fun taking the test! My scores: 200 ms : 10/10 100 ms: 10/10 50 ms: 9/10 (my cat came asking for food and i got a bit distracted, lol!! ) 30 ms: 9/10 20 ms: 9/10 10 ms: 5/10 5 ms: 2/10 54/70 Not a musician :)
When I studied neurology my lecturer mentioned a lot of experiments with squid axons and how fast an impulse travelled along a neuron. That's probably the limiting factor here in terms of what you can perceive. I remember when I used to use the awful DirectX audio drivers before ASIO that my playing was terrible because the latency was always over 20ms. Anything under 10ms, the latency is pretty much imperceptible. Between 10 and 20ms is bad but you can adjust to it.
I just tried it and got at least 9 out of 10 until 10ms with 6 and 5ms with 7. I was not expecting to do that well at all. I tapped along with it which might be why I did slightly better but I'm not sure.
Playing music by theory alone, is like rock climbing with shoes. They’ll get you to the rock but, make it tougher to climb if you can’t feel or get into a groove. Only after you abandon the shoes (theory) and bare your soles (feeling) can you reach the top!
This is the video I've needed to watch for years. I remember the first time I really felt microrhythm was listening to Eminem's "Shake That". It felt SO laid back and I didn't know why! I later learned about Dilla beats and that led me to realize that music in general uses microrhythm and that music is not meant to be played perfectly on the click. It makes 100% sense why in jazz you hear that backing tracks and iRealPro can be so harmful and that nothing beats playing with real people. I have since then learned to stop intellectualizing the music I listen to and instead just listen to it for what it is.
I think the reason why you can feel the difference in the music even if you cant accurately determine the subtle shift in the tests is simply because in the music those differences are now in context instead of isolated like they are in the test. For example, the ghost notes of the quantized drum track at the end get totally lost in the mix due to the quantization, where in the original they stand out way more.
Was definitely guessing by the end of the test. Guitarist/bassist here, I've been using a DAW to record most days for a few years, and I feel like that's what's made me more sensitive to hearing micro time. I can definitely hear things that I couldn't 5 years ago. I wonder how much being able to see performances up against the DAW grid helped improve my perception.
Funny enough, when I first pressed play, my computer audio was still getting routed through my bluetooth sound system, and the video and audio signals were about 50 ms delayed.
Finally somebody talks about this! I’ve been trying to find the language to describe the micro rhythmic characteristics of individual drummers’ grooves forever. It would be amazing if you could talk about Stewart Copeland in this respect. It’s no secret that he plays with tempo a lot but I’ve always wondered what the characteristics of his subdivisions are that make his style so unique.
Very interesting thing - just made the test. At least I'm proud to not be that wrong and mix "too early" with "too late", even not at 5ms. At the end I discovered (as a classical trained percussionist), "too early" does sound bad to me in every case. "In time" and "too late" from 20ms to 5ms were harder because "too late" can sound very good, kind of end phrasing
When taking the test, I noticed that my performance was way better at 20 ms than at 30 ms, not only with regards to my accuracy (9/10 vs 4/10), but also with regards to how confident I was in my answers. It might have been random, but it would be interesting to know how common such variable accuracy is, and also what might cause it. Anyway, thanks for putting this together, Shawn! I really enjoy experiments like this one.
This is very interesting. I found I did better than I thought I would. I also found that my internal time sensor rushes. On all but 1 that I got wrong I was ahead. I thought it was on time when it was early, and thought it was late when it was on time.
Very good Shawn! Have you spoken yet about a collab study with The Music Lab? It would be cool to see a more fully vetted study of these results! Also, as described in a response to Adam Neely's Tweet about your efforts here, Timbre will probably have an effect which would be good to control for, as well as the notion of whether all tests within a 10 question group would benefit from taking them off grid from question to question (I couldn't help but note that they were all at approx. 4 beats pause before the next batch of strikes), or whether the gap between changes our perception. Anyway, Kudos!
Yep, timbre probably does matter! Would be interesting to know how this effects pitched instruments. For example, does a sine wave register in our brain faster than, say a distorted chuggy guitar, where the first 50 ms (or whatever it is) of a note is 'pitchless' attack, before you actually hear the note? All kinds of weird variables to explore with this.
Hi bro, i realized one thing a while ago. When you listen only with your ears, and there is a limit to what you can perceive, that limit increases quite apparently and obviously when you are the one controlling the sound. 😄 - i noticed this a long time ago, certain things like barely adding reverb, or whatever effect or edit to a groove like adding some swing, You can adjust it with a physical knob, and are able to perceive even more minute variances since the tactile input is also basically being processed in your brain. Kind of letting you know that something is happening, and your audio sense is almost extrapolating the tiny change. it really is quite a strange thing to experience. and a very weird phenomena. Maybe you could do a video about that some time? That would be amazing! Basically manipulating your perception based on changing sensation inputs, sound, tactile, light.. Like the human is an instrument lol. And measure what we can perceive using different methods of input. And then there is also training vs. the human limit. Like if someone had more extensive training to be able to identify abnormally small variations normally people can not distinguish. like training an ai network. (which is literally what our brain is, and neural networks are modeled after lol) basically we are a computer just accepting input and synthesizing data inside our brain. xD
Great video! Would be interesting to see if earlier notes are harder to detect than later notes, and if that changes with the amount of delay/anticipation. Also interested to see if most mistakes are in one direction or another or equally distributed (i.e., more people think early beats are on time and on time beats are late or the other way around). Sorry, my PhD student brain is racing with experimental possibilities!
The 50--15ms flam window you cited seems pretty relevant, considering that people could still hear offsets with greater-than-random accuracy all the way down to 20ms. Presumably, a flam any tighter than 15ms would start to blend too much.
Took the quiz as a never professional musician but on again, off again music student for many years and I did a lot better than I thought I would! I tended to skew later (guessing on time for early, and late for on time) more than earlier. It wasn't until 20ms that my answers were very off (late when it was actually early x2) but I still got 6/10... there was still a lot of guesswork and I might have just gotten lucky. For 10ms and 5ms I was 1/3 +/- error bars. 10, 10, 9, 8, 6, 4, 2
my score: 56/70 100% on 200ms down to 30ms; 8/10 on 20ms, 4/10 on 10ms-5ms pretty proud of this, since i'm mostly just a math nerd who dabbles in music theory but doesn't practice
Just wanna say I appreciate you not ascribing value to quantized vs in quantized and stressed the importance of these semi-conscious decisions in music.
After a certain point since rhythm and pitch are just sound waves, wouldn't you at a certain point just hear pitch? So if you subdivide at fast enough tempo you would hear a pitch instead of a "groove"
As a listener, I’ve never been able to hear the difference between quantized and unquantized beats. Hell, I can barely tell the difference between triplet swing and quintuplet swing.
Did way better than I thought I would, 10/10 down to 20ms (with one miss on 30ms) 6/10 on 10ms and 4/10 on 5ms I don't play any instruments but I occasionally mess around with a DAW.
So I find this in particular very interesting: there is a drop at around 30 ms and then a slight increase towards 20 ms.(Myself included, I also had a big difference between 30ms {6/10} and 20ms {8/10}) This had me thinking: would 30ms be somehow unnatural and guide our brains willingly into choosing the wrong option? So as that if one were to make music with 30ms beats one would be in a constant state of unsureness and distress which would make the music feel very off? :)
Pro producer tip from deadmau5's masterclass: in a standard house beat where the kick and snare both play on beats 2 and 4, if you delay the snare slightly by a few milliseconds like in this video, it lets the kick come through nice and clean without sidechaining, and people won't hear the delay.
thats such good tip man! i just tried it on a pitched snare in a future bass track, keeping the dry snare with the kick making both the kick and the snare feel completely separate. It really highlights the snare transient allowing it not to get washed out by the entry of the kick. This is acc so useful thnx a lot
I took the test and was super fascinated by my results. I had 8 years of band between 1999 and 2007 just from 5th to 12th grade, but since then I've regularly played rhythm games, to a relatively high level of difficulty. I do a lot of drumming on my desk, lap, etc as a result and do consume a good amount of music content online, but don't currently play any instruments. I got a perfect score up through 30 ms, and got 80% on 20 and 10. I got 50% on 5 ms, but there was an answer I was insisting was late, and the button would not let me select late for some reason, LOL. I don't know if that would have been correct though. I got an adjacent answer wrong, so I'm not sure which one it was. Still, I do think it's fascinating, since I guess I am somewhat rhythmically self taught and didn't expect to do as well as I did.
Hey ! Very interesting video as always ! Just being picky here but I've been trained to analyse statistical set like yours and I know it's not really important yet I'd like to precise a couple of things. When you plotted the accuracy compared to frequency you sure did obtain points and a curve but you cannot just by looking at this, as you did, tell whether or not what you see, in terms of differences, is significant. Maybe between two points the accuracy drops by 5% but is it really significant ? If for each point you show standard deviation bars, you will show for each point how much all the measurements scatters from the mean. Therefore by observing the overlap of standard deviations bars between points, you can say : - overlapping = "same thing" so no significant differences between the two accuracies for the two frequencies - non-overlapping = different groups, the variation of accuracy is significant . Hope I was clear :D
Yes, of course. Lots of flaws in the experiment. This was not a proper scientific study, though I'd love to see someone do one on this. Hundreds of people have taken the test since yesterday (soon to be thousands probably), maybe more data will help determine the amount of variation for a given measurement.
Huh, fascinating. I was perfect down to 50 ms, and then started missing late hits but I was perfect on detecting early hits down to 10 ms. (And that's not just post-facto random chance -- I was much more certain about hearing early hits than whether one was on time or late.)
everything can be notated, you just have to be ready to break out the 128ths and 96-tuplets
I know only all the 69-tuplets from EMCproduction :/
@JM Coulon I'm sure theres ways to do the same purposefully with modern software
@JM Coulon to be honest my area of expertise is composing using synths and sample libraries so for me it would be ideal to be able to have that level of control and precise instruction with a session musician if I needed it, but in practice people aren't really able to execute that, and it really isn't fun for them to try
a decent tactic that does sometimes work is to program a loop that has the microrythm, dynamics, and feel you want, and then say "do your best to match the feel of this" and give it to them in the headphones instead of a click.
Everything can be notated, but not always in a way that is purposeful for a musician to read. If your Dilla-groove is looking like a Ferneyhough piece something has gone wrong.
im glad this channel has turned into drummer adam neely, love it love it love it
I mean, they both get to use Sungazer in their videos 😏
aaaaand kiss
Subscribed after reading this
Repetition legitimizes.
His presentations have the same professionalism seen in Nahre Sol's work. The way they speak is very similar.
Don't wanna jinx it, but I'm so thankful that traditional copyright laws don't include drum melodies or grooves, so we can analyze drum legends to our heart's content :)
unfortunately it does. once i uploaded drum groove played by me, from song that has drums-only intro, and got claim
Really appreciate your pronunciation of Ernst Heinrich Weber.
@@rzufig961 did he ever live in NY?
@@quinreimer5906 pretty sure he did at some point
Yeah I thought so too
@@rzufig961 oh right that explains it. i was also like whoa, what just happened
check out his screencaps (e.g. 8:55), turns out even his computer is in German.
Fun fact: dogs apparently have almost perfect sense on rhythm. Saw it on QI episode. They trained a dog like 118bpm means going for a walk, 120bpm means food. Now if I could only trained my dog to bark when the drummer strats in the wrong tempo! Animal metronome.
drummer: *drags*
dog: not quite my tempo
Animals as (band) leaders?!
I hope you do a follow up on how this affects the "chemistry" between a drummer and a bass player. As a bass player I've played with a lot of drummers and sometimes it is easy to play together and sometimes it is really hard. Over time I've started to feel that this is mostly an issue of having or lacking a shared understanding of microtime. After all, if a beat is slightly behind the quantized position, this could mean two things: tempo modulation or microtime and both of these would imply a different time at which the next beat should occur. The same holds in reverse as well - I might play a note just slightly behind a bass drum hit as a microtime choice and the drummer might try to adjust for me dragging.
Another possible follow up would be something on how to use quantized and unquantized grooves as a means to generate contrast.
Agreed. When a drummer/bass player finds their counterpart that they fit with, you just know. Can't put your finger on it, you just know the two of you got it.
Have a read into musical entrainment, there’s some academics doing pretty fascinating research on isochronous rhythms & dynamic systems. Some use the social/metaphorical application, beyond the standard “physics” parameters, such as how well the players know eachother etc. Always excited to see people exploring these changes in phase relationship
I believe the micro timing perception is likely going to be relative to the tempo in those tests. By increasing (or decreasing) the tempo, a fixed shift of the 5th note will represent a larger (or smaller) portion of the beat interval. This would be an interesting controlled variable to test.
Wow, that's a quite important control measure not thought of in this experiment. I guess the frequence/ pitch of the note also plays a role. Maybe a high-pitched "beep" that most metronoms give you will cause a better distinction - and headaches in all participants :D
Exactly. It's not the same to detect 200 ms differences in 0.5 seconds than in 5 seconds.
I’m proud to call myself an unprofessional musician that can distinguish a difference of 30 msec groove
I get unreasonably excited every time Shawn uploads a video. Man, you do such a great job breaking these concepts done in such a way that it is a joy to learn.
100% agreed!
The fact that I was like "hmm feels like 10-15 ms early" on that last example makes me far more excited than it probably should.
I remember asking you about something related to this in chat not too long ago and I thought of you again when I saw this vid... and now youre here!
@@martinskrebergene :)
Same, editing my own bass playing for tracking has left me very hyper aware of my timing.
Excited enough to practice and try to incorporate this?
Bro SAME! I think it's probably because you've been grinding specific timings at 60Hz for OoT speedrunning, because I've been grinding my Super Smash Bros Melee (you'd enjoy the process of getting good at melee, I think) timings, and I was thinking, "oh, that felt about a frame early" and it was right in that neighborhood. So fucking sick to realize that your brain is capable of processing timing windows that quickly. Anyway, you're dope, keep up the good work my man.
This is incredibly valuable for anyone who records themselves practicing, thank you so much!
I’m a composer that started on a non rhythmic instrument, and the fact that I now have an answer to “how do I write a groove” has made my month
I can feel my drummer brain expanding from this video
As a self taught guitarist (15 years) and newbie bass player (4 months) I really expected to do terrible, but my scores were consistently 8 and 9 until 10ms. 10 and 5 were rough... this is all as fascinating as people's pitch perceptions. I love studies about this stuff, great video!
Excellent stuff! I reckon Shawn has already read the famous Finnish study from 2015 analyzing Jeff Porcaro's timing: "Fluctuations of Hi-Hat Timing and Dynamics in a Virtuoso Drum Track of a Popular Music Recording".
7:22 The KidPix "oh no!"
Very neat study, I wonder what would happen if you performed this survey on rhythm game players as well?
True, that would be interesting. The hitwindow on higher difficulties can get pretty low. I don't know about other games but in osu! OD10 is +-19.5ms, so pretty much at the threshold that can be consistently perceived, according to Shawn's data here. Although you'd have to survey some really high level rhythm gamers for them to compete with drummers with 10+ years experience ^^
In Rock Band, there were some very fast notes that I couldn't hit until I turned the calibration to the nearest 5-10 milliseconds. Anything outside that threshold and the game wouldn't register it.
Yo what the fuck KidPix that’s a deep memory what the fuckkkk
I'm a student musician that plays violin, viola, drumset, and marimba. Been playing for around 7 years.
I got all correct up until 30 ms
Got 1 wrong on 30 ms
Got 5 wrong on 20 ms
Got 6 wrong on 10 ms
And then 4 wrong on 5 ms
My overall score was 54/70, although luck was on my side for the last one :p
Percussion, haven't played in a few years (apartments, man), played for about 13 years basically consistently until I was ~21, always in a student context. All correct through 50, 2 wrong on 30, 8 wrong on 20, 9 on 10, and 4 on 5.
i've pretty much only ever used DAWs, but played a bit of keyboard in the past.
i got all correct up to 3 ms
1 wrong on 30 ms
2 wrong on 20 ms
3 wrong on 10 ms
but only 4 right on 5 ms
my total score was 58/70
maybe listening to lots of speedcore (~ 300+ BPM) has something to do with it?
I have no musical experience apart from attempting to play the first "Smoke on the water" riff and singing in the shower. Got 57/70, made one mistake in 30ms and 20ms, half wrong in 10ms and 6 wrong in 5ms. I did re-listen and check it for 2-3 times though, not sure if it's considered cheating.
You should do a similar experiment with rhythm gamers. I was in school band class for 6 years, and have been playing rhythm-based games (DDR, Beatmania IIDX) for another 20. Quantized beats are literally the objective. I got 100% all the way through 30ms just like the professional drummers, as did a number of other rhythm game fans. I think you're definitely on to something with that range of perception being trainable.
The unquantized beat at 14:00 made me bob my head while the quantized version did not. It's very subtle, but you do feel it.
Hey Shawn,
I felt like the comparison of the "flat and lifeless" sound of the quantized version vs the original version of Steve Jordan's recording was really unfair to the sound of a quantized beat. Wouldn't a large part of that be the negative effects of attempting to quantize raw audio? I feel like in order to have a significantly more fair comparison we should have either two raw audio recordings where one is played with the 20ms delayed ghost notes and the other closer to being in the pocket/quantized; or perhaps using 2 digital drum tracks where we can control with much more accuracy exactly where each hit is placed. I loved the rest of the video and have been a fan for a long time! Keep it up!
I agree... not trying make a case against quantized beats (though I understand if people interpreted it that way). that segment was simply meant to be "check out this example of microrhythm!" anyway, I actually did the MIDI comparison at the time of the original experiment (didn't include in vid only because of time constraints). here are some examples you can check out: drive.google.com/drive/folders/1h7eQcaDVxBvRba2Z2--TlVjTB8Mo9vQO?usp=sharing
It's quantized.. and then swinged! ;)
Dude, props to your data interpretation. Most people who don't have a background in experimental sciences don't know how to properly interpret and - more importantly - criticise their data.
This would be interesting to have in a mobile app. Then you could see how your time awareness improves the more you work on it.
I've actually been working on a mobile metronome app, which I could modify to do this.
Would love to see producers taking this test
Very very nice. Did the test and found out I can hang with fellow professional drummers!
me too :p
I'm very happy with my results!
9/10 - 200ms
10/10 - 100ms
9/10 - 50ms
10/10 - 30ms
5/10 - 20ms
6/10 - 10ms
5/10 - 5ms
After the 30ms I can't spot the diference. Can't tell I really heard (or even feel) if the beat was late or early.
I'm 31 now, I had drums class when I was 14 and again with 16. Them I start playing acoustic guitar, had classes for 1 year and stoped. I still play, but only for fun and with no consistent what so ever. What I do is watch your channel and other channels related to music stuff and listen to music on a daily basis. I don't have a good ear for pitches, but I have a good ear for volumes, don't know if this help to understand my results.
I really like this video and your channel!
(I don't speak english but I tryed my best to type this)
I have very similar results and story... I am curious - do you drum with your hands on your legs, or desk, or anything a lot? I find it very difficult to keep still, and drum with my hands a lot like this and have done since I last played drums about 20 years ago. I wonder if this has helped my score?
@@jonzenrael Hey
Yes, I do, but I wouldn't say a lot. Never have a hard time trying to stop the drumming. I last played drums about 4 years ago
Your research made me think about two interesting ideas that might explain the drop off around 20ms besides just the limit of human perception.
1) You should try to group your dataset by early/late samples. I have a hunch that early samples will be a lot harder to detect than late ones. I expect this from your choice of sample. While a very short sound, a snare hit is not a perfectly "point like" sound. It smears in time for at least 5ms. I think that even if you play it early (-20ms), since there is still noise at the moment we expect (0ms), we still perceive it as on time.
2) At the speed of sound in air ar 20C, 1ms is 34cm. I'm sure you know how people who hear themselves talking with a delay will have a tendency to slow down. What if this is what's happening in that last performance? You mentioned how the delay was more pronounced on the kick, it also happens to be the furthest away from your hear. This is a very subtle effect, and its generally known that under 5ms of latency in recording is not perceptible. But what if it is, only subconsciously? In this case, we're talking 2-3ms at best... what if we include nerve response time? Assuming that pain/reflexive reponses are the same speed as conscious programmed triggers (like when you try to follow a beat and you consciously know when you should move your arms to hit the next note at the right time), well, those are at about 15ms... together thats... awfully close to the 20ms delay we're seeing here.
I feel like there's a lot of interesting questions to answer around this topic. More research is needed.
THIS is what happens when you hear a drummer play a couple measures and realize he or she is a good drummer. Like you said "even if we can't consciously percieve them, we can feel that they are there". Sometimes can't tell exactly what makes a good drummer (or any musician), but I think microrhythm is an important spark added to it.
Thanks for this! and please, do not erase the test, I would like to make a video about it, giving you the credit, of course!
This is great that you bring this up. We producers usually play around with pushing some hits forward or backwards by samples. Basically very low ms. I had a test in production school where I had to listen to a test like this. We also analyzed songs that use this technique to make a song feel fast and jittery or dragged and relaxed. I had a kinda easy way to feel that 10ms faster note because it felt stressed to me. Deliberate dragging and rushing is extremely important to make a song have the right kind of feel like you showed. VERY overlooked sadly.. Max Martin is a MASTER production technique. Even with midi drums.
Drumming ABSOLUTELY changed how I "felt" music as I played it. I'm totally unsurprised we're well represented in this niche music skillset.... and was surprised by how confidently I could see myself taking this test... until you brought up drummers 🤣
I love everything about this video 🤙🏻 Everything is well explained and I like how you even went into the ms necessary to identify a difference. Microrhythms are the key to studying the center of mass of rhythms and seeing what feelings are produced when those small shifts in timing happen. There's all sorts of fun stuff there. Great work!
It’s kinda of a big relief to hear that I’m not the only one consistently rushing or lagging by 10-20 ms. I’ve often manually quantized the feeling out of my jams. 😂
This just came up in my feed and I remember taking the test when it came out. My thoughts this time is if hearing these different flams is teachable, the group to test would be snare players from the top drum corps and marching bands. One thing that dci produced during the golden years was methods for educating students. Certainly there are instructors out there with trucks for eliminating the flams between players.
Yesterday i discovered you. You are my new master! Cheers from Germany!
Incredible my guy. Thank you SO much. This is exactly the video I needed as I’m diving deep into groove
Everyone go watch Anderson Paak’s NPR Tiny Desk Concert. The second and fourth songs feature him applying some microrhythmic style on the backbeats!
Yes, i do microrhythm. I write metal music with this kind of idea in mind. Really cool video man!
Stats per genre would be particularly interesting. Great vid, man
Played in a drumline in middle school for 3 years.
Results:
9/10 for 30 ms
10/10 for 20 ms
5/10 for 10 ms
5/10 for 5 ms
Quite happy with the results. Can I microrhythm? Apparently yes!
I’m not a musician, but you made it easy to understand and appreciate the topic. (Kudos on presenting the caveats of your research too!) The quantized vs original comparison was awesome!!! Thank you!
I actually noticed that i was way more confident in my ability to tell when the note was early, even at 20ms, but at that point it was more of a feeling based guess of if it was late or in time, that seems like an interesting thing to note
I was trying to put a finger on this exact thing a few weeks ago when I was trying to explain to a friend why I love Anderson Paak’s drumming so much, it’s not flashy but his feel is so good.
Shawn, I've been following you since watching your videos with Adam Neely, and later you inspired me to learn Japanese (still in that journey)! Thank you so much!
One thing I noticed about the microrythm test is that I changed HOW I was listening throughout the test. At first, I was focusing my attention on the attack of each beat to determine if the 5th hit was early/on time/late. However, somewhere around the 30-20ms mark, I realized I was better off listening to the entire sound envelope, paying close attention to the decay of each hit. That really improved my score: although I scored a 5/10 on the 30ms test, I scored an 8/10 on the 20ms!
Something to think about! Keep up the great work,.
In the beginning of the video I was really suprised that most people could feel a difference of 20ms. But then again, the way we tell the direction a sound is coming from in part has to do with the different times at which our left and right ears recieve that sound. So there are actually a few aspects of our perception of sound that occur at really small time scales.
Not saying the topic of this video and the thing about direction are related. But they're both messing with me because I've never thought of perception as being that quick I guess
Extremely interesting experiment.
200ms, 100ms and 50ms maxed out, then it went downhill from there. Interesting fact - at 30 ms, I made 3 mistakes, all of which were assuming that something is in time when it wasn't. It took me an additional tier to start seeing ghosts.
10ms and 5ms are just pure random as expected.
ps. primary music school (6 years) graduate on the clarinet. Nowadays musically active as a hobby.
New sub here, great video! Just worked through the test- 100% score until the 10/5ms rounds whereby my brain apparently switched off and I scored 1 and 4 points respectively.
Listening 1-4: 10 of 10
Listening 5: 7 of 10
Listening 6-7: 5 of 10
Guitar is my main instrument.
Before the test, I thought that finding the early ones might be easier, because both playing behind or on the beat feels "good," while early beats feel uncomfortable to me. But actually I did all kinds of mistakes :D
One weird analysis I draw from this is the use of drummers even in the age of DAWs. Don't get me wrong. I started out with drums and am now a keyboardist and after watching tonnes of your and Adam's videos, have also increased my musicality and knowledge of theory and other things. The point is, like you said in the end that Jordan sir didn't consciously think before adding those milliseconds worth of delay, he just did it naturally, unconsciously. And that's what can happen naturally only. You can't program such a natural feel into DAWs. So no matter how better DAWs might get, Sungazer still needs you man! Just kidding, hope you get the point.
Big fan sir!
Thanks for this contribution Shawn. Is't beyond interest and amusement, it is showing a human skill of adapting (like i call it) the universal swing. My abilities (after 40 years of drumming) end beyond 30, maybe 20 ms...... which explains my BIIIIIG tolerance for timing fluctuations. The whole topic explains roughly why my fellow musicians over the years - well-known and unknown - differ so much in the acceptance of timing variabilities and the evaluation of them, together with the judgment of musicality in relation to timing.......
On the production side, with regards to creating reverb, the old rule of thumb is that up to 20ms of time delta is perceived as reverb, after 20ms it is an echo.
I also find it interesting that 20ms is roughly the maximum of input/output/roundtrip lag that any musician, professional or not, may find acceptable when playing their instrument - which is mostly of course a problem in the digital world only. And that's why we all will eventually use ASIO and external interfaces.
I've also found that the more experienced I get, the more sensitive I am to being bothered by any such lag. When I was younger, I had no problem creating music with a keyboard/output roundtrip that definitely had closer to 30ms of lag. Now it is out of the question, even 10ms is only barely tolerable, momentarily, before I want to throw the PC out of the window.
Also, for the results of your study, I think the actual distinction is people who routinely perform against a metronome and people who don't, rather than professional or non-professional musicians per se.
But, of course, a professional is much more likely to have spent countless, hundreds and thousands of hours perfecting their ability to play against a click, otherwise they're probably not getting hired for anything.
Anyway, cool vid!
That's so crazy I've been thinking about this topic all week because I notice even amongst world class drummers some seem to have better time accuracy than others
Did the test and the results pretty much followed the same patterns as the other professionals with 15 years plus experience. Really interesting. There must be a point where it becomes almost impossible to perceive the difference, as you suggested in this video. Nice work, always enjoy your videos.
I read a study for my college course on microtiming and it’s impact on Groove and against everything what many music TH-camrs say the study showed that perfect quantized recordings received a very high groove rating sometimes even higher than the original recordings. Although I should say that the „grooviest“ versions of the recordings did always have some microtiming.
The fact that I got any correct @ or below 50 ms is pretty crazy. One issue is the short time between examples. I was still clicking the answer when the next one started.
Please note that choosing "in time" for all answers for 20ms, 10ms, and 5ms (which I did because I couldn't hear any error) gives you a score of 50%, 40%, and 40% on each of those, _not_ the 33% expected from a randomly chosen result, so the survey results should possibly be analyzed/interpreted differently.
Also, as someone who scored 70% on 30ms and as noted couldn't hear 20ms of error, I *also* cannot hear any difference between the quantized and unquantized Steve Jordan performance, not even as a subconcious "just sounds better" sort of thing. (Ditto the Rick Beato video where he also quantized a drum performance.) I am curious to what degree this is true for others, and how that breaks down between professional musicians and non-. And if it is different between them, what this says about the process of music creation when the creators have aesthetic opinions the listeners do not.
daru jones is amazing on subtle offbeat rhythms. got 53/70 on the test myself :)
Very cool video and had a lot of fun taking the test!
My scores:
200 ms : 10/10
100 ms: 10/10
50 ms: 9/10 (my cat came asking for food and i got a bit distracted, lol!! )
30 ms: 9/10
20 ms: 9/10
10 ms: 5/10
5 ms: 2/10
54/70
Not a musician :)
OT: it's noticeable that you took the challenge of pronouncing Ernst Heinrich Weber seriously. That's awesome!
When I studied neurology my lecturer mentioned a lot of experiments with squid axons and how fast an impulse travelled along a neuron. That's probably the limiting factor here in terms of what you can perceive. I remember when I used to use the awful DirectX audio drivers before ASIO that my playing was terrible because the latency was always over 20ms. Anything under 10ms, the latency is pretty much imperceptible. Between 10 and 20ms is bad but you can adjust to it.
I just tried it and got at least 9 out of 10 until 10ms with 6 and 5ms with 7. I was not expecting to do that well at all. I tapped along with it which might be why I did slightly better but I'm not sure.
Playing music by theory alone, is like rock climbing with shoes. They’ll get you to the rock but, make it tougher to climb if you can’t feel or get into a groove.
Only after you abandon the shoes (theory) and bare your soles (feeling) can you reach the top!
your videos are just great! thanks for putting so much work in them and of course thanks a lot for sharing!
This is the video I've needed to watch for years. I remember the first time I really felt microrhythm was listening to Eminem's "Shake That". It felt SO laid back and I didn't know why! I later learned about Dilla beats and that led me to realize that music in general uses microrhythm and that music is not meant to be played perfectly on the click. It makes 100% sense why in jazz you hear that backing tracks and iRealPro can be so harmful and that nothing beats playing with real people. I have since then learned to stop intellectualizing the music I listen to and instead just listen to it for what it is.
I think the reason why you can feel the difference in the music even if you cant accurately determine the subtle shift in the tests is simply because in the music those differences are now in context instead of isolated like they are in the test.
For example, the ghost notes of the quantized drum track at the end get totally lost in the mix due to the quantization, where in the original they stand out way more.
Wow great introductory musical lesson on butterfly effect
Loved this type of video, thank you
Was definitely guessing by the end of the test.
Guitarist/bassist here, I've been using a DAW to record most days for a few years, and I feel like that's what's made me more sensitive to hearing micro time. I can definitely hear things that I couldn't 5 years ago. I wonder how much being able to see performances up against the DAW grid helped improve my perception.
Funny enough, when I first pressed play, my computer audio was still getting routed through my bluetooth sound system, and the video and audio signals were about 50 ms delayed.
Finally somebody talks about this! I’ve been trying to find the language to describe the micro rhythmic characteristics of individual drummers’ grooves forever. It would be amazing if you could talk about Stewart Copeland in this respect. It’s no secret that he plays with tempo a lot but I’ve always wondered what the characteristics of his subdivisions are that make his style so unique.
Very interesting thing - just made the test. At least I'm proud to not be that wrong and mix "too early" with "too late", even not at 5ms. At the end I discovered (as a classical trained percussionist), "too early" does sound bad to me in every case. "In time" and "too late" from 20ms to 5ms were harder because "too late" can sound very good, kind of end phrasing
When taking the test, I noticed that my performance was way better at 20 ms than at 30 ms, not only with regards to my accuracy (9/10 vs 4/10), but also with regards to how confident I was in my answers. It might have been random, but it would be interesting to know how common such variable accuracy is, and also what might cause it. Anyway, thanks for putting this together, Shawn! I really enjoy experiments like this one.
This is very interesting. I found I did better than I thought I would. I also found that my internal time sensor rushes. On all but 1 that I got wrong I was ahead. I thought it was on time when it was early, and thought it was late when it was on time.
Very good Shawn! Have you spoken yet about a collab study with The Music Lab? It would be cool to see a more fully vetted study of these results! Also, as described in a response to Adam Neely's Tweet about your efforts here, Timbre will probably have an effect which would be good to control for, as well as the notion of whether all tests within a 10 question group would benefit from taking them off grid from question to question (I couldn't help but note that they were all at approx. 4 beats pause before the next batch of strikes), or whether the gap between changes our perception. Anyway, Kudos!
Yep, timbre probably does matter! Would be interesting to know how this effects pitched instruments. For example, does a sine wave register in our brain faster than, say a distorted chuggy guitar, where the first 50 ms (or whatever it is) of a note is 'pitchless' attack, before you actually hear the note? All kinds of weird variables to explore with this.
You gotta mention dilla!! He brought it to mainstream detached from any genre aside hip hop
Man that was super interesting
62! (10, 10, 10, 9, 9, 8, 6) One happy outlier, presumably not guided by chance at all.
Hi bro, i realized one thing a while ago. When you listen only with your ears, and there is a limit to what you can perceive, that limit increases quite apparently and obviously when you are the one controlling the sound. 😄
- i noticed this a long time ago, certain things like barely adding reverb, or whatever effect or edit to a groove like adding some swing, You can adjust it with a physical knob, and are able to perceive even more minute variances since the tactile input is also basically being processed in your brain. Kind of letting you know that something is happening, and your audio sense is almost extrapolating the tiny change. it really is quite a strange thing to experience. and a very weird phenomena. Maybe you could do a video about that some time? That would be amazing!
Basically manipulating your perception based on changing sensation inputs, sound, tactile, light.. Like the human is an instrument lol. And measure what we can perceive using different methods of input. And then there is also training vs. the human limit. Like if someone had more extensive training to be able to identify abnormally small variations normally people can not distinguish. like training an ai network. (which is literally what our brain is, and neural networks are modeled after lol) basically we are a computer just accepting input and synthesizing data inside our brain. xD
This is some quality content.
Great video Shawn, would love to try the test.
Great video! Would be interesting to see if earlier notes are harder to detect than later notes, and if that changes with the amount of delay/anticipation. Also interested to see if most mistakes are in one direction or another or equally distributed (i.e., more people think early beats are on time and on time beats are late or the other way around). Sorry, my PhD student brain is racing with experimental possibilities!
Excellent video as always
The 50--15ms flam window you cited seems pretty relevant, considering that people could still hear offsets with greater-than-random accuracy all the way down to 20ms. Presumably, a flam any tighter than 15ms would start to blend too much.
Took the quiz as a never professional musician but on again, off again music student for many years and I did a lot better than I thought I would!
I tended to skew later (guessing on time for early, and late for on time) more than earlier. It wasn't until 20ms that my answers were very off (late when it was actually early x2) but I still got 6/10... there was still a lot of guesswork and I might have just gotten lucky. For 10ms and 5ms I was 1/3 +/- error bars.
10, 10, 9, 8, 6, 4, 2
Great video Shawn. Very interesting subject, So educational. Thank you!
my score: 56/70
100% on 200ms down to 30ms; 8/10 on 20ms, 4/10 on 10ms-5ms
pretty proud of this, since i'm mostly just a math nerd who dabbles in music theory but doesn't practice
Just wanna say I appreciate you not ascribing value to quantized vs in quantized and stressed the importance of these semi-conscious decisions in music.
Was not expecting that perfect pronunciation of Ernst Heinrich Weber
took the test hope it help 48 / 70. subscribed .
Great content as always
Incredible, this is really eye-opening!
After a certain point since rhythm and pitch are just sound waves, wouldn't you at a certain point just hear pitch? So if you subdivide at fast enough tempo you would hear a pitch instead of a "groove"
yes and there's an adam neely video about it
In theory yes, the hard part is playing that fast :)
As a listener, I’ve never been able to hear the difference between quantized and unquantized beats. Hell, I can barely tell the difference between triplet swing and quintuplet swing.
I'm amazed I pulled off 59/70 on this, getting near perfect until 10ms, where it dropped to a solid 60% 😎
Did way better than I thought I would, 10/10 down to 20ms (with one miss on 30ms)
6/10 on 10ms and 4/10 on 5ms
I don't play any instruments but I occasionally mess around with a DAW.
So I find this in particular very interesting: there is a drop at around 30 ms and then a slight increase towards 20 ms.(Myself included, I also had a big difference between 30ms {6/10} and 20ms {8/10}) This had me thinking: would 30ms be somehow unnatural and guide our brains willingly into choosing the wrong option? So as that if one were to make music with 30ms beats one would be in a constant state of unsureness and distress which would make the music feel very off? :)
Pro producer tip from deadmau5's masterclass: in a standard house beat where the kick and snare both play on beats 2 and 4, if you delay the snare slightly by a few milliseconds like in this video, it lets the kick come through nice and clean without sidechaining, and people won't hear the delay.
thats such good tip man! i just tried it on a pitched snare in a future bass track, keeping the dry snare with the kick making both the kick and the snare feel completely separate. It really highlights the snare transient allowing it not to get washed out by the entry of the kick. This is acc so useful thnx a lot
I took the test and was super fascinated by my results. I had 8 years of band between 1999 and 2007 just from 5th to 12th grade, but since then I've regularly played rhythm games, to a relatively high level of difficulty. I do a lot of drumming on my desk, lap, etc as a result and do consume a good amount of music content online, but don't currently play any instruments.
I got a perfect score up through 30 ms, and got 80% on 20 and 10. I got 50% on 5 ms, but there was an answer I was insisting was late, and the button would not let me select late for some reason, LOL. I don't know if that would have been correct though. I got an adjacent answer wrong, so I'm not sure which one it was.
Still, I do think it's fascinating, since I guess I am somewhat rhythmically self taught and didn't expect to do as well as I did.
Hey ! Very interesting video as always ! Just being picky here but I've been trained to analyse statistical set like yours and I know it's not really important yet I'd like to precise a couple of things. When you plotted the accuracy compared to frequency you sure did obtain points and a curve but you cannot just by looking at this, as you did, tell whether or not what you see, in terms of differences, is significant. Maybe between two points the accuracy drops by 5% but is it really significant ? If for each point you show standard deviation bars, you will show for each point how much all the measurements scatters from the mean. Therefore by observing the overlap of standard deviations bars between points, you can say : - overlapping = "same thing" so no significant differences between the two accuracies for the two frequencies - non-overlapping = different groups, the variation of accuracy is significant .
Hope I was clear :D
Yes, of course. Lots of flaws in the experiment. This was not a proper scientific study, though I'd love to see someone do one on this. Hundreds of people have taken the test since yesterday (soon to be thousands probably), maybe more data will help determine the amount of variation for a given measurement.
@@ShawnCrowder I'll be happy to help if you want :D
i wonder if it has something to do with processing speech, i think being able to feel like the groove of the way a person talks is pretty important eh
rhythm games brought me here, but this makes so much fall into place
Huh, fascinating. I was perfect down to 50 ms, and then started missing late hits but I was perfect on detecting early hits down to 10 ms. (And that's not just post-facto random chance -- I was much more certain about hearing early hits than whether one was on time or late.)