Comment by jdietrich | Hacker News Reader

Comment by jdietrich | Hacker News Reader

Parent Full thread

>I’ve always wondered, is there a better way to annotate music?

Not really. Western staff notation is heavily optimised for intelligibility by an expert user. It's not very intuitive, it barely works at all for non-western pitch systems, but a trained musician can read, understand and reproduce complex scores in real time. For the primary users of western musical notation, that is an essential, non-negotiable feature that negates every disadvantage.

Western notation leans heavily on both visual pattern recognition and an understanding of music theory. It can be read in chunks, like written natural language. When you're reading this comment, you're not parsing individual letters, you're recognising whole words and phrases. It's the same for music notation - a skilled user sees chords, progressions and scale fragments rather than individual notes. They are parsing the dots and lines as meaningful musical information, not purely abstract pitch and duration data.

It's not a particularly precise or machine-readable system, but it's not supposed to be. Where we need machine-readability, we have MIDI and OSC - they're totally unreadable by humans, but they convey a vastly greater range of data at vastly greater precision than any traditional notation system.

Notation encodes as much detail as is musically relevant and no more. For example, we typically only use eight levels of loudness, from ppp to fff. It's not really useful for the composer or the performer to encode very precise loudness information, because human musicians can't reproduce that level of precision.

Western notation is kind of amazing. It encodes timing by horizontal space, pitch distance by vertical space, harmonic distance by accidentals complexity, and has a sort of symmetry around A 220 in clefs, and the octave which is the most basic interval happens to be the height of one staff.

I can see why it's not easy to give up.

Where it suffers is in notating timbre, so you get things like the orchestral score which isn't really all musical information but also leaks implementation details.

I've always been fascinated by the possibility of a chromatic western notation though, which would require way less learning but would therefore not centre around the C major scale.

Imagine it as a piano with all alternating black and white notes, no gaps.

Now there are 6 white and 6 black notes per octave.

You would only have to learn two positions to have learned all scales and all chords (like sliding moving a bar chord on a guitar).

I don't know that it would be better. It might be soulless but it also reduces the redundancy of learning harmony and theory and suddenly jumping up a key is as simple as physically moving up a key.

Look into chromatic keyboards like the Jankó keyboard or the Isomorphic keyboard.

You'd also need different keyboards and notation if you start considering music not built around the octave.

Thanks for that I hadn't seen the Jankó keyboard, although the purely physical fix doesn't deal with my bigger frustration that the accidentals and quirks of scale centric notation requires a huge investment in theory learning to learn Western music in more than one key... Having taught theory from scratch a couple of times as well as learned it, it's been a lifelong frustration that the same thing written a different way can be massively more complex even though the semi-tonal pattern is identical. I'd go as far as to say that the language and notation around scales actually obfuscates the patterns and then when you really understand the circle of 5ths you realise it's just a veneer and your mind can just translate it.

Really interesting reference. Will definitely read more into Jankó.

Interesting almost every pro musician I know uses the chord numbers to convey this information orally more than they use the note names. And when I did degree in jazz it was always focused on the numbers.

Musicians clearly understand the patterns, but it's beginners that I'm concerned with. After decades of study / practice most systems are workable.

> Where it suffers is in notating timbre

There are a great many annotations to individual notes: lines, accents, staccato dots, slurs and brackets, dynamic markings etc. etc. that indicate timbre, and you've always got the fallback of just writing out what you want, and it's even acceptable to not even use Italian these days!

We could have that same intelligibility for less-expert users if we didn't have all the legacy gunk. Of course, the expert users have a vested interest in having a more-exclusive club. Nearly all of them would reject anything new.

A simple improvement is to give each of the 12 common pitches within the octave a distinct position on the staff. This eliminates accidentals from typical music, leaving them only needed for quarter-tone compositions. Sight-transposition becomes trivial, possibly within reach of beginners.

> A simple improvement is to give each of the 12 common pitches within the octave a distinct position on the staff.

This doesn't work - the staff becomes too large, or the registral range it covers becomes way too small. The current staff has evolved to be readable without expressly counting the lines or spaces, via visual subitizing. If you add more lines, you lose this property and sight-reading becomes infeasible. Plus, diatonic pitch is actually meaningful in a musical sense, and it's useful to have a notation that references it. That's why we use for instance C, D, E, F, G and not <0>, <2>, <4>, <5>, <7>.

(This also describes the very real problems w/ Klavarskribo, BTW.)

Key signatures and accidentals have a practical benefit in indicating the tonal centre. All pitches are not equal, at least not at any given moment. Sharps and flats look like a kludgy hack, but they provide clear visual hints of highly relevant musical information.

There are alternatives to just having a huge number of lines. Western music notation already has one: the grand staff has a gap in the middle. Simply put a gap with a single ledger line between each octave and you get a system that looks an awful lot like the current system. It's like spreading two octaves over a grand staff, but with a more modest gap in the middle.

So you can still have sight-reading. You also get fully arbitrary sight-transposition, which today is only possible for the very most elite musicians.

But you want each key to look different because physical instruments are all tied to a key. If every transposition looks the same then you've just turned a visual transposition problem into a physical transposition problem.

Physical instruments aren't really tied to a key.

Transposing instruments only even exist because sight-transposing is so difficult. There is nothing fundamental about an instrument that makes it transposing. We only use that kludge because of the difficulty sight-transposing.

Suppose you do have a transposing instrument though. It's in Bb, and you'd like to play music written for an instrument in F. Very few musicians can do that without hesitation.

The same applies when there is a desire to shift the entire piece of music up or down. Perhaps you are using a hecklephone instead of a lupophone, so you shift everything up. Perhaps you just lost a trumpet player with an unusually high range, and thus need to shift everything down. It would be great to be able to expect all players of ordinary skill to be able to do this without hesitation.

BTW, I'm aware that one needs to avoid troublesome notes and things like impossible trombone glissandos. This doesn't change the general truth of the above.

> Western music notation already has one: the grand staff has a gap in the middle. Simply put a gap with a single ledger line between each octave and you get a system that looks an awful lot like the current system.

Except that's not how a multi-staff partition works. Each staff has a clef that determines it's offset; they can be the same or wildly different. They also usually signify different "voices" or instruments, further encoding musical nuance beyond pitch and tone.

You say it doesn't work but it's been used in a variety of ways such as singing video games (eg. SingStar) and music trackers [1], and it works just fine.

[1] https://en.wikipedia.org/wiki/Music_tracker

Working in a game where mistakes don't really matter and working for professional music production where mistakes and restarts cost $$$$$$$ are nowhere close to being the same thing.

Sure. And orchestras can hire teams of page turners so that people can play more than 3 measures at a time without stopping. And conductors can . . . I don't even know what a staff that large would look like for a full orchestral score. I guess you can just give the conductor a screen and have it scroll continuously. And then the orchestras can hire an ops team to support the scrolling screens.

Starts to sound a lot like the same kind of thinking that gave us service oriented architecture and a new javascript framework every week. Just hire shitloads of more people to make sure it all works when it needs to. Because it's easier for beginners! In fact, with this new notation system that simplifies everything so much, you could probably fire all the older more expensive musicians in the orchestra and hire a bunch of juniors for a lot less money!

They use screens and color cues for performances where a live orchestra plays back the soundtrack of a film in sync with the projection. Professional musicians are a lot more open to alternative representations and technology integration then you might be aware.

The screens are for the conductor, not the musicians. They're following the conductor's direction and don't need to know what cues they're playing to, nor can they take their eyes off their score.

There simply isn't the money in the budget to bugger about with clever ideas in a sync session. Merely having a session with an orchestra is already a luxury for many productions, because most composers can knock out a reasonably convincing orchestral soundtrack in VSL at a fraction of the cost.

An orchestra is an extraordinarily efficient musical machine if they're doing what they were trained to do. If you want innovation, don't hire an orchestra - experiments get very expensive very quickly when you're paying ~$50 a minute for the privilege.

https://www.youtube.com/watch?v=HaWGWQJVVBI

No, the height increase is only by a factor of 12/7, which is about an extra 71%. Compensating for this, there is a width reduction caused by the loss of accidentals. There is also no need to waste space on unused octaves, so there might not be any height increase at all for some pieces of music.

Trivial transposition is a highly desirable property.

>Of course, the expert users have a vested interest in having a more-exclusive club.

I am an almost entirely self-taught jazz musician. I learned to read notation begrudgingly, assuming it to be an archaic hurdle put in place by snobbish classical musicians. My view of classical musicians has not changed, but my view of notation has.

A lot of people think that programming languages are a similar archaic hurdle and set out to build a visual programming environment that will be easy for beginners to use. Somewhere along that journey, they invariably realise that a) syntax is one of the easiest parts of programming, b) there is no good visual metaphor for recursion and c) the Deutsch limit is a cruel master. You can do simple scripting with a box and arrow diagram, you can teach kids to blink an LED, but you can't write a compiler.

The advantages of stave notation (and the disadvantages of alternative systems) are non-obvious, but they are very real. Critiques of notation usually sound like "command lines are stupid, everything should be a GUI" - they're not wrong per se, they just miss the point.

> ...there is no good visual metaphor for recursion...

Now, now. You're on a site that's literally named after one!

I'm taking classical piano lessons currently, and I have much the same reaction. I don't love all the strict rules imposed by western music, but I do love the notation and theory, because it generalized in interesting ways across tunings and notation.

Just like I learned C (for example) before imagining my own languages, or how I've learned many libraries before writing my own, I also appreciate the structure of classical notation for what it is.

>c) the Deutsch limit is a cruel master.

I generally despise visual programming but MaxMSP does this pretty well. You can nest components in an almost object-oriented fashion. Prior textual programming experience doesn't massively help learning the language though, it's a whole different paradigm and it's def not easy for beginners.

Do you know if the nestability thing applies to puredata (open source Max/MSP)?

I haven't touched puredata yet but C&G's Organelle is looking really amazing right now.

Yes, you can create subpatches (http://write.flossmanuals.net/pure-data/subpatches/) and abstractions (http://write.flossmanuals.net/pure-data/abstractions/).

I'm happy with standard notation when playing the piano or something that can only play one note at a time (e.g. trumpet), but I still can't get my brain to read it fast enough when playing the guitar.

I know tab notation exists to help with this, but it loses much of the detail encoded in standard notation - note lengths being the most obvious example. There have been attempts to create a hybrid notation and there are books with tab printed below the stave(s), but neither work well.

If anyone's found any alternative notations that work well for guitar, I'd love to see them.

I've played from books that had tabs with note stems, beams and flags attached to the numbers with the same meaning as in standard notation (this can't distinguish between a filled and unfilled note-head, but in guitar music one rarely needs to notate anything longer than a crotchet/quarter-note so it's not really an issue). I never really learned standard notation, but I do know some of the basics and found it useful.

But this would make it much harder to sight-sing typical melodies, eg one whose notes are all diatonic to G major.

Reason: a melody whose notes are all diatonic to G is almost indistinguishable from one whose notes are all diatonic to C in your system while in the current one they are quite distinct. Furthermore, your system would make it very easy to count chromatic distance between notes, but harder to count the diatonic distance (or the scale degree a note lies on).

It's interesting how staunchly people will defend "standard Western musical notation" but I agree with you - from a design point of view it looks like notation originally designed to match a piano keyboard which has then had many years of extra features inelegantly bolted on to it. It works well enough for the keys of C or Am but gets progressively more ridiculous the further you depart from those keys. And transposition is a mess.

Some people seem to think that standard musical notation is "right" and you can't possibly have any alternative notation and yet this is clearly false. There are at least two or three alternative forms of musical notation in wide use today. Most guitarists use tablature notation because it's better suited to that instrument - standard musical notation is less useful on guitar since it doesn't clearly capture the different ways that a single chord can be played. So it makes sense to have a special notation for that instrument.

Secondly chord charts are used by musicians everywhere because they're very concise at conveying the chord structure of music and often that's all the information you need. Standard notation is just way too much information in that case.

The third, less commonly seen form of musical notation is just like the one you've described. It's seen in video games like SingStar[1] and in modern music composition software such as Ableton Live[2] and others. It's used so often in music composition software because as you've noted it's very intuitive and far less cumbersome than standard musical notation. Transpositions become trivial and everything just makes more sense.

[1] https://www.gamesmen.com.au/media/catalog/product/cache/1/im...

[2] http://whereisdarran.com/wp-content/uploads/2014/02/Screen-S...

> It works well enough for the keys of C or Am but gets progressively more ridiculous the further you depart from those keys.

If you're in something other than C or Am, you're supposed to use key signatures to point that out and clean up the notation. C/Am is just the default.

"Standard musical notation is ambiguous when used on guitar", actually it isn't if you mark strings and fingerings as needed. This is widespread practice, each instrument has its own "tweaks" on the basic system. Standard notation in particular is quite hard to work with on guitar, and one can see why so many people just rely on tablature - but ambiguous it's not.

Chord charts may be "concise" but they don't actually convey harmonic structure very well, not least because of the multitude of informal conventions and the ambiguity that these create; figured bass notation (while admittedly not very well known) does a far better job of it overall. For example, "walking" bass lines become immediately apparent in thoroughbass.

My point was that these notations exist for a purpose and are popular. They do and they are.

Why would tablature be so popular on guitar if standard musical was overwhelmingly better? Tablature is intuitive, expressive and concise when used with guitars. Just as standard musical notation has its strengths so does tablature. Ok, tablature's poor at expressing rhythm but standard musical notation's terrible for transposition. So they're all just different tools with different strengths.

The one thing that "standard notation is overwhelmingly better" at is notating music, and making it understandable as such. Some people care about this enough that they'll put in the effort to use standard notation even on guitar; some just want to focus on how each specific piece should be played on the instrument, for which tab is quite optimal. It's even quite common to have a format where both are notated, on separate staves.

(For that matter, standard musical notation is not even that terrible for transposition once you understand how key signatures work, via the cycle of fifths. It's a bit tedious, but it absolutely can be done; one can even learn to transpose directly on sight. And good tablature notation will of course include accurate info about rhythm, there are quite easy ways of doing this.)

I’d argue that tablature is terrible for transposition as well in most cases, sure you can just +/- N semitones from each fingering in the ideal case but that breaks down as soon as you have voicings or melodies with open strings and need to transpose down e.g. a half step. Capo at 11 gives a very different character.

>actually it isn't if you mark strings and fingerings as needed

You mean, if you add tablature above each chord (or, effectively, transcribe that information via numbers, etc)? :)

When the parent said standard notation, he meant the one without this augmentation, for which it wasn't quite optimized.

>Standard notation in particular is quite hard to work with on guitar

That's the point - that it's not the use case the notation primarily was designed for. Guess we all agree here, don't we?

Accidentals and key signatures are a feature, not a bug. There's a reason we don't just write everything in C major.

Reply on news.ycombinator.com