What is tightness and how do we measure it?

In designing BeatBalance, one of my main goals has been to help drummers (including myself!) tighten up their timing.

Instinctively, we know tight playing when we hear it.

But to give accurate feedback on something, we need to be able to measure it. And to measure something you need first to have a clear concept of what that thing is.

So what is tightness? I’m going to start from as close to first principles as possible here so please excuse me if anything I say seems facile or like I’m stating the obvious!

Whatever tightness is, it’s a property of musical timing. There are lots of other important elements of playing - pitch, dynamics and timbre for example - but they don’t play into tightness except inasmuch as they vary in time.

A single drum hit can sound good, be well played in other ways, but whether or not it’s tight inherently depends on its relationship to something else - to the other notes in the pattern you’re playing, to a metronome click, to other musicians.

So what’s the appropriate reference point?

Let’s have a look at a few different ideas.

“Tightness is how on the grid you’re playing”

Modern recording software provides us with one way we could define tightness. On the arrange page of every modern recording application there’s a grid of evenly spaced subdivisions, and it’s now standard practice to quantize a lot of playing to that grid, so everything is perfectly ‘in time’. There’s an easy reference grid to see where the beat is - so it’s tempting to conclude that everything that’s not on the grid must be out of time. Metronomes also provide a regular pulse (more on that later).

This type of perfectly quantized rhythm is extremely familiar to us nowadays, but as with the Kraftwerk track above it can often sound a little robotic (even if it’s in a odd way).

But does that mean that non-metronomic, evenly spaced timing is inherently sloppy and untight?

Tight bands don’t always play exactly on the grid. This famously sampled James Brown track breathes and speeds up and slows down, and a metronome set going in sync with it would drift out very quickly.

As an even more extreme example, Indonesian gamelan music has a complex, fluid sense of time, with the music speeding up and slowing down, but it depends on highly coordinated, tight playing.

So it’s better to talk about tightness less in terms of the relationship of playing to the grid or ‘on the beat’ but rather in a way which ensures that what you’re playing is well coordinated with the other things that are happening.

David Byrne of Talking Heads put it well in How Music Works:

"[Tightness] doesn’t actually mean that everyone plays exactly to the beat; it means that everyone plays together….. lurches and hesitations are internalized through performance, and after a while everyone knows when they’ll happen…. agreed-upon imperfections are what give a performance character, and eventually the listener recognizes that it´s the very thing that makes a band or singer distinctive."

“Tightness is playing consistently together with other elements of the music”

This makes a lot more sense. We feel that a piece of music is tightly played when all the elements which should sound simultaneously *do so*.

So as drummers, what can we do to improve the overall tightness of the music we’re playing in terms of our own playing? Or in other words, what is tightness for us as individuals?

If we’re trying to make the elements of the music gel, then we can obviously only control our own playing, not any of the rest of what’s going on. If we’re playing alongside other musicians they can adapt their timing to match ours (and vice versa) - note that David Byrne talks about agreed-upon imperfections. And if we’re all over the place timing-wise it will be very difficult for the other players to hit that moving target.

The better we can control where each of our hits is falling, the more we’re able to adjust ourselves to fit others’ playing and widen the ‘pocket’ of the groove. If the bass player is a little

bit late on the offbeat, we can move our kicks back a little bit more consistently to match.

So from an individual perspective, tightness is about intention - the ability to place what we’re playing precisely *when* we intend it to happen.

“Tightness is intentional control of timing”

This is the definition that I think makes most sense, and the one I use to guide design choices for BeatBalance. When exactly any particular note actually falls is irrelevant, as long as the person who put it there intended to do so. This also makes sense of the apparent paradox of ‘loose’ playing - a lot of great drummers have a loose, flowing feeling to their playing which is often contrasted with ‘stiff’ or ‘metronomic’ playing. But these players still have a huge amount of intentional control over the timing of what they’re playing.

Highly skilled players even in extremely rhythmically complex and metrically irregular music such as Malian djembe drumming are able to synchronise their playing with one another just as well as musicians playing more metrically regular types of music.

So why do we use metronomes at all?

If playing ‘on the grid’ isn’t the measure of tightness or correct timing, why use a metronome to practise and improve your timing? Won’t practising to a metronome just make your playing more rigid and robotic?

Plenty of drummers have developed an excellent sense of rhythmic coordination and play very tightly without ever having used a metronome. So using one is clearly not essential for developing great timing.

In order to learn anything, however, you need feedback of some kind - whether it’s your own judgment, that of others or some external measuring device. If you can’t discriminate between a bad and a good performance then improving will be impossible.

Music teachers of course can provide immensely high quality and useful feedback. Playing with other people provides excellent feedback, as does listening to other music and being immersed in musical culture.

But the reality is that for a lot of people the majority of instrument practice time will be alone and in this context anyone faces the problem that playing music and simultaneously judging how well you’re doing is a challenging task. Especially challenging is tightness, because as we’ve explored tightness is on the one hand inherently relational, but on the other deeply about intentionality. If we’re just playing on our own then to a large extent we’re ‘marking our own homework’. So it’s useful to have ways of getting feedback which don’t involve people.

A metronome is useful because it provides an external target to aim at, and allows us to more easily judge whether or not we’re placing our strokes when we intend.

A metronome is a great feedback device. Developing skill in music is as much about developing perception as manual skill, and comparing your playing to a consistent yardstick allows you to see weaknesses or inconsistencies in your playing more easily. Identifying weaknesses and inconsistencies quickly and addressing them helps you progress faster, so the better and more immediate the feedback you get, the faster you’ll be able to improve. I want BeatBalance to provide more helpful information to analyse playing and address problem areas.

At the moment, the app provides only a relatively simple selection of rhythms. One of my priorities for the next stage of development is to expand the range of rhythms available - different time signatures and rhythmic subdivisions (triplets, quintuplets etc) as well as some more esoteric non-isochronous rhythmic options which I will discuss in a later post.

To help players improve their tightness, I wanted to come up with a simple, accurate but easy-to-understand way of measuring tightness and translating that into a useful, intuitive score.

Scoring tightness

Although the app gives you breakdowns and graphs showing your timing, it can be difficult to get an intuitive grasp on what these mean. When we’re discussing tightness we’re talking about timing on the order of milliseconds (or tens of milliseconds), and although this is a time scale which is perceptible it’s hard to get an intuitive grasp on what it means in the way that most people have an intuitive understanding of what it means for something to last a second or an hour.

To contextualise this information I wanted to break the information on tightness down in a way which is accurate but easy to understand intuitively.

The most common way to quantify tightness in drum apps is to use a percentage system - usually a percentage showing how many hits you got within a certain window of time around the target. This was the system I used in early versions of BeatBalance but there are a couple of problems with it which emerged during testing.

Firstly, a percentage score implies that perfection is attainable - if you’ve scored 100% it’s simply not possible to play any better. One of the things that I’m not sure about here (and I don’t think any studies have been done to answer the question) is quite what the limits of human tightness are.

Related to this is the fact that I want BeatBalance to be a useful tool for players of all levels, and it may well be possible for even highly skilled players to continue to improve their timing over a long period.

The other problem is that a percentage system presents a bit of a ‘cliff-edge’ problem - a strike within the allowable window is worth something, but no more than any other, and a strike outside is worth nothing, and no less. This means that subjectively quite similar performances can end up getting very different percentage scores.

As a result the app currently uses a different system. The system scores tightness out of 1000, and works by calculating a measure of the spread of the overall timing performance. The more narrowly grouped the performance around the target time, the higher the score. A score of 1000 corresponds to a performance with no perceptibly out of time elements at all.

To get slightly technical for a moment for those who are interested, the score is calculated as a ratio of the standard deviation of the performance to the standard deviation of a normal distribution where 3 standard deviations corresponds to the limit of what is perceptibly out of time (which itself we calculate as either 3% of the inter-beat interval at that tempo or 15ms, whichever is the larger).

There’s also a star rating to make this all a little easier to understand - a score of 1000 is not ‘perfect’ but it’s intended to be good enough under all circumstances, so this is given a 5-star rating. 800 points is a very good performance and gets 4 stars, 600 3 stars and so on down.

I’m by no means certain this is the best of all possible systems so I welcome feedback and discussion!