Mostly good info in there, but I see a few problems:
> There is also no E#. It just doesn't exist. Get used to it. Those are the 12 notes.
Technically not true. There is an E#, it's called F. If you are playing in F# major[1], the scale has an E# in it.
> Musicians never play notes one at a time. They always use chords.
Most lead guitar parts are not chords. Typically playing in a band context will create a chord from multiple instruments, but that isn't always the case.
If you want to get a bit more wild, though, a lot of solo music still contains chords by implication, and can be very easily harmonically analyzed. I don't know how popular solo works are in the 20th century since atonality flung the music locomotive gleefully through the "Bridge Out Ahead" sign and into the canyon [1], but if we discount that it would certainly include almost all Western solo music.
[1]: Yes, it's an opinion. Yes, it's an informed opinion, and implying in one way or another that I just must Not Get It will not faze me. I do Get It, and that's why I find it distasteful.
I used to think that, but when I had to study atonal music, it became clear to me how much of an impact it had on film music and how it shows up in fragments in pop music.
Then, I took a class on electronic music composition. For my own project in the spirit of the avant garde, I wanted to challenge myself to write something "musical" without resorting to the traditional concepts of pitch-based melody and harmony or quantized rhythm. It turns out, there are many other levers you can play with to create the effect of an musical arc. Not to pat myself on the back, but I think the end result is evocative in a similar sense as tonal music. If I had the time, I would do more.
If we want to play "wave the credentials" at each other, I minored in music composition, and my objection is a great deal philosophically deeper than "I don't generally like how atonal music sounds". There's a parallel world where atonal music is still explored but I wouldn't characterize it as "flinging music off a cliff". Atonality was merely when the cliff-jumping occurred.
I have neither the time nor the desire to fully explain my point in an HN post, but I would note that the music community itself has been asking questions in this direction in the last decade or so, if you want to follow up on it, as the academic music community asks themselves exactly how they got to where they are now, so disconnected from the rest of the world at large and so utterly irrelevant to almost anything musically occurring today. (Not just pop music; that is structurally inevitably insipid, really, so who cares, but irrelevant to anything other than their own very little world.)
also lo-fi sound chip music, but chord progressions may be emulated with fast arpeggiato (sequential instead of parallel chord notes, the name is from the harp, which exhibits this effect). There's probably a limit at which slow speed we don't notice notes as connected, but violin is surely fast enough. Speach pretty much seems to rely on an aliasing effect like that, too, when frequency clusters (iirc) are enough to identify words in speech analysis. So order would not matter as much as chord, if I am not mistaken. But then I might be as things get more complex than single progressions. I wonder if interleaved arp progressions have been explored.
I learned this by asking a friend in the CouchSurfing meetup who brings a guitar or a roll-out piano. And as I say, I can't play any instrument myself.
Maybe if I want to learn, I should choose an instrument that uses single notes, it might be easier.
Regarding singing though, my ex-girlfriend told me to sing more quietly in church because it sounds so bad.
> Maybe if I want to learn, I should choose an instrument that uses single notes, it might be easier.
It's not just limited to bass/lead guitarists either - finger style acoustic guitarists don't play just chords. OK, often the fingerings are chord patterns, or arpeggiated chords, but that's not always the case. Checkout the title track from Bensusan's album, "Intuite", that is mostly just a single note melody: https://www.youtube.com/watch?v=di1qQsGcaIc (it would appear the ending is cut off this track here, but it gives you the idea).
That's nuts. That's like asking a beginning programmer to stop coding because their code is garbage. The only way to be a great singer is to be a bad singer first.
> I should choose an instrument that uses single notes, it might be easier.
I'd say the easiest / most immediately rewarding way is to pick up a guitar or similar (banjo / uke / banjolele / mandolin etc) and learn a few chords. Although people who can do that aren't exactly in demand so there's a case for learning the onde martineau or glass harmonica or something - that way if you get decent at it you can actually get paid to do it.
Or any instrument that isn't tuned to 12-tone equal temperament. In just intonation there are no enharmonic notes, for just one example. C# isn't always the same thing as Db!
And that's without mentioning oddities like most varieties of bagpipes or the hurdy-gurdy, where a constant drone is present. It doesn't exactly form chords with the other notes, but it's still more than one note being played at a time!
That’s a fair point. But still, the relationship between the root, 5th, and major third are very much connected to the fact that they are specific multiples of that root frequency, and are the loudest overtones typically audible when the note is played. The root is not totally different from chords built on it.
I meant to add more to my top comment. I think the author does fairly well at their mission despite potential nit picks like that. It is really cool to see somebody learning and being excited about putting all the pieces together. The beautiful thing about music is you can go deeper and deeper as much as you want, and there are various competing ideas about what it is and how to talk about. Programming has a similar kind of depth to it.
The fact that the same thing has 2 names, but then the same name (e.g. C) can mean 3 different things (note, chord, key) really confuses me. I'm a beginner (can't play any instrument), so this is just about trying to cover the basics and generate a machine learning dataset. If I weren't so confused already, I might try to learn an instrument. It seems to be really time-consuming to practice though, and I'd need a teacher to listen and correct my mistakes. I wouldn't notice when I get it wrong if I try to teach myself.
I'd suggest starting with the tin whistle. They're very cheap, very easy (you can learn the basics in a day or two, a week at most to get your first tune down), and any Irish session will either have someone willing to teach or someone who knows where you can get a teacher. The Session[1] is an excellent resource for it.
As for which whistle to buy, get one in the key of D. A Walton's little black should be about $6, one of their slightly better brass whistles around $11, and a Generation nickel D is about $12. If you really want to splurge Susato makes a very nice high D for $50 (their Kildare S series), but it's quite loud and more for if you are playing before an audience. The most expensive whistles around are probably Burke's, at $230 for a high D. So even if you end up collecting Burke whistles obsessively and getting one in every key he makes you'll still be out less money than buying some high-end guitars!
The one disadvantage is that they're diatonic instruments. That means you can play in one key, and only one key, without cross-fingering the whistle.
For tuning, I recommend either an electronic tuner or the program Flutini[2], assuming you get a tunable whistle (the cheap ones aren't and it probably won't matter much).
Non-musician who taught himself all this stuff here, I feel your pain.
That said, Abb isn't another name for G - it's a distinct thing which, when played on typical musical instruments, is played the same way as a G. Basically note names to note frequencies is a many-to-one mapping.
If you want to understand the hairy details start by googling just vs equal temperament. But I went down that rabbit hole, and it hasn't made my procedural music any better, so I wouldn't really recommend it.
It's not a fixed mapping. It's more like a function that produces a mapping. It depends on the selection of:
- A pitch standard, like A440, which anchors one arbitrarily chosen note to a specific frequency.
- A tuning system, which defines the frequency ratios between notes.
Almost all modern music assumes _A440_ and _12-tone equal temperament_. In 12TET, adjacent flats and sharps map to the same frequency. It's said that they are _enharmonic_. But there are many other systems where this isn't true. In _just intonation_, notes are defined by simple frequency ratios, and there is no enharmonicity at all. Every number of sharps and flats map to different frequencies.
Even this is not quite right. For keyboard instruments, that have a fixed set of pitches that notes map to, you have to choose a temperament, a fixed set of pitches for each note (unless you're on a clavichord, where you can adjust it a bit by how hard you strike the key). Fretted instruments may or may not be tempered. Guitars are, since the frets on it limit the pitch range of a given note, though it is still a range. Viols aren't, because the frets are soft gut and used to brighten the sound, so the pitch range is much wider for a given note.
For string or wind instruments playing without keyboards, there is no temperament. Each note represents a range of possible pitches, and what pitch you choose depends on the notes around it. For example, if you are repeating or holding a note, you don't want it to change pitch. But if the harmony around that note is changing, it may be in tune at one point in time, then out of tune at a later one. String quartet playing involves a great deal of working out compromises about what pitch each note should have.
Good point! I admittedly was writing this off as "performance details" of music written in a modern 12-tone paradigm, but you're definitely right that plenty of music is written specifically for ensembles with continuous pitch with those nuances in mind. See also, a cappella vocal music, particularly barbershop. Even for guitar, some players tune to favor particular chords shapes sounding maximally consonant, even though the frets are designed for 12-TET.
its also about chord spellings within a context- how a musician, especially in a string quartet, thinks about how to play the note based on how it is spelled (eg is it the third, fifth, seventh or some extension of a chord?) the spelling tips you off to its context based on the key and other factors you feel but don't have time to contemplate when sight-reading.
Something I think a lot of people don't realize when starting out in music: you don't actually have to know how to play an instrument to make music. You don't even need music theory.
Knowing a little about how instruments work and learning some music theory can help tremendously, but you can actually just start poking around in a MIDI roll or tracker and go a long way.
Unless you're really into a specific instrument, starting off with any modestly priced keyboard with pressure sensitivity and MIDI out (USB-based if it's modern) will let you have fun with synths and sampled instruments you'd probably never be able to afford.
If you really want to learn a physical instrument that's easy to pick up, you can get a $20 plastic ocarina and find lots of tutorials on YouTube.
In theory it should run in any Chrome/Chromium browser, but I've received several reports of rhythatom failing to play. Would appreciate any help or ideas.
There doesn't seem to be any indication of major/minor. One way to interpret that is "always use the diatonic." Which means only use notes in the key, which means chords (1,4,5) are major while (2,3,6) are minor. However songs can have non-diatonic chords.
If you look at Rhyathatom, it defaults to the well-worn 1645 progression - the 6 is explicitly minor, which makes it diatonic. Try making the 6 major and you get a different animal - kind of sinister! That's a non-diatonic chord.
Nice work! Am I crazy or are the pulldowns for accidentals not doing anything?
> There doesn't seem to be any indication of major/minor.
There's a line in there about capital/lowercase of roman numerals not being important, by which I hope the author meant out of scope of the article. So I think they were explicitly just enumerating the chords of the diatonic scale.
I got this error the first time but worked fine on subsequent refreshes:
An AudioContext in a cross origin iframe must be created or resumed from a user gesture to enable audio output.
G.snd.init @ (index):98
init @ (index):464
A for effort, but as a programmer and musician, simultaneously makes me a little sad as an example of the all-too-frequent phenomenon of programmers trying to model a domain without spending enough time actually learning the nuances of the domain first.
Nuances? Try basics. Worse, the author makes a number of general statements about music that I would have known to be wrong as a 6-year old with 1 year of experience in playing piano.
Who am I kidding? This is the age of crapping some mangled mis-information out onto a blog or web page in the hope of raising one's profile and securing a job with the next hot startup creating $10k robots delivering $5 pizzas or whatever.
The author clearly and good-humoredly declares that he comes with no knowledge of music or music theory at all, and is clearly documenting his process of coming to grips with enough of those basics you may have known as a 6 year old with 1 year of experience, and is attempting to do so in order to do something interesting. So perhaps 'wrongness' is far less important to this than 'interestingness' and, though I am a multi-instrumentalist myself and know music well, I'd like to thank him for attempting to learn what he needed to to attempt this interesting thing. I really enjoyed the read and intend to have a little play with the data.
The whole music investigation was a side-project, compared to my main efforts on Pingtype (a Chinese learning site). https://pingtype.github.io
I got interested because my friends at the KHOP prayer room gave me the song sheets in PDF format. They have the chords written, and the key on the top right. I thought it would be interesting to figure out which songs go well together so I could make some playlists. Even though I can't play music, I listen a LOT - all day in the office, and while I bike to work.
The fact this side project got more upvotes than my language-learning interests, and your encouragement that the data might be useful, is a sign to me that I should keep on with this and maybe learn to play. There's a piano in the house I rent, but I don't know what to do with it. It seems like now might be a good time to start learning.
Ok, perhaps you are fine with letting little gems like "Musicians never play notes one at a time. They always use chords." slide, but I'm not. If I can't trust the author about the basics and things that I know about, how can I trust their reasoning or thoughts on more complicated matters and items that are outside my own domain? I cannot, and hence don't read further because there is an opportunity cost in doing so, and there are many other interesting things I could be reading or otherwise doing with my time.
Honestly I just feel like you may have wielded judgement too hastily at the expense of your sense of wonder and shared joy in someone exploring territory that is well trodden only to relatively few. Very few people have a solid grasp of musical theory. For most people, scaling what may to you be barely the foothills of knowledge is a great and challenging experience, and I think it should be celebrated.
And sure, perhaps you and I can easily distinguish the subtle difference between 'Musicians always use chords' (obviously wrong) and 'Musicians always work, individually or together, within a key, which is an abstracted but well-defined set of tones that can be combined in time series or simultaneously to produce an inexplicably pleasant experience in the listener, by some neurological magic that has never been adequately explained' (non-obviously correct) but y'know what, I actually am willing to let little gems like that slide because they're adequate approximations of the truth that allow good folks to proceed on a journey that may one day lead them to, just maybe, have the arrogance required to belittle the uninitiated.
For me, the issue was that the writing style, like most articles written by programmers, passed off every thing as fact. Things like "they have the same chord progression" in section 2.a is patently false. This kind of writing is misleading to others, especially newcomers (blind leading the blind)
I do agree though, that to do somethimg, learn something, have something almost correct is better than having learned and done nothing, but as Bill Evans points out in the youtube video about the learning process, it's much better to start with something simple, concrete, and which you know as true then build from there, than start approximating (incorrectly) lots of things at once. So it's a case by case basis, but it's not merely gratuitous arrogance that motivates people to call other people out on false statements.
> Very few people have a solid grasp of musical theory.
This is just like saying that very few people in the world have a solid grasp of computer programming... Do you know that there are hundreds of university departments in the world where the only thing people do is, wait for it... having a solid grasp of musical theory!?
Not everyone if the world is trying to create a computer model of music. The minimum a person with this interest can do is to contact a specialist in music theory (of whom there are thousands as I mentioned).
Yes, exactly. The well temperated musical tuning [0] is "good enough". it's an approximation. Some pitches for notes sound almost the same and that's why they today are represented with the same key on a piano.
> For some reason I don't totally understand, some have 2 names (e.g. C#=Db, D#=Eb, F#=Gb, A#=Bb).
"There's also 7th chords, Augmented chords (+), and Diminished (º) chords. As far as we're concerned, there's a lookup table that lets us turn these chords into notes."
This is nice as a diary of some hacking, but much of the musical commentary is misleading. Basically the author has enumerated all the unique ways of selecting the numbers 1..7 four times with replacement, and then done various mechanical translations on the results. Ostensibly this is done to enumerate all possible four-bar chord progressions, though in practice the results are highly redundant (i.e. many results are musically identical) and incomplete (many possible chords aren't included - e.g. secondary chords).
Then rather than distributing the scripts the author presumably used to do the enumerating and translating, they are distributing the outputs of the scripts, as a huge tarball full of midi files. Not sure what the motivation here was.
The way I see it, most people prefer to take existing data (e.g. Kaggle datasets) and build machine learning tools on that, instead of having to compile everything from source and make their own data.
In the Parsons Code section, if I'm reading correctly, the chords are conceived along an up-down axis. However chords do not really have that relation. Going from 1maj to 5maj can be up a fifth or down a fourth.
To illustrate this, first listen to the default 1645 progression on rhythatom. Then change the octave of row 2 (4maj chord) from +0 to +1. Can you hear its similar role in the progression, even though it's now stressed by being up an octave?
I talked about this with my friend Lenard, because I was confused why the Parsons code would be different for the same chord progression.
He then explained that I should move the key, and when I transposed the MIDI table, the Parsons codes all became the same.
I then asked him, is it possible to move up an octave for just one part of a chord progression? He said yes, it's possible, but people don't do that often. Normally people wait until the end of a chord progression to change octave.
Talking about going up or down an octave here is just nomenclature. Musically speaking, changes in a chord progression don't go up or down - this is because roman numerals refer to pitch classes (like C, E, G), not to specific pitches (like C4, E5, G3).
The matter of which notes to play for a given chord is called "voicing", and it's orthogonal to anything mentioned in your article.
Another correction:
> The notes in "Cm" are C,D#,G.
It's actually C Eb G. The difference is that D# is an augmented 2 while Eb is a minor 3. They sound the same but they have totally different functions.
Yes, but only with pitch bends. For example the Great Highland Bagpipe is tuned to a just intonation scale, which commonly approximates A Mixolydian (neither a major nor minor scale). (It's actually 3 pentatonic scales overlaid, but let's not get into that...)
I would invite the OP to read a basic book on music theory, any good undergrad-level text should do, before stating some of the "facts" he is making, which are just untrue.
For one thing, the author is confusing physical pitches with musical notes. C double-sharp exists and is enharmonic to D, for example, because the frequencies are the same. But these are very different notes in a musical context.
There are many examples of this in the article that would be cleared up by a basic understanding of music theory that all music students complete, many of whom do this in piano lessons growing up long before graduating from highschool.
I admire the effort, but I often see people analyze music from a perspective that is not actually that of a musician, and there are always huge gaps in the understanding that needn't be there if they sought a basic knowledge of the matter.
Example:
> For some reason I don't totally understand, some have 2 names (e.g. C#=Db, D#=Eb, F#=Gb, A#=Bb).
Fair enough, you don't understand this, so that should be a clue that more knowledge is necessary before making this claim:
> There is also no E#. It just doesn't exist. Get used to it. Those are the 12 notes.
Incidentally, it goes much further than this, and for good reasons. F is the same pitch as E# and G-double-flat, but all three are distinct musical notes which serve different purposes.
Putting aside the rest of the nits here, there are two things that stick out in my mind. First of all, what he calls 4-chord songs are really more "4-bar songs" (which given his restriction of one chord per bar end up being at most 4-chord songs), but we should be clear about the distinction between the tones being played and the duration of those tones.
Second, I don't know if this is an artifact of my fluidsynth setup or not, but when I translate his midi files into WAV files the 6 and 7 chords are from the octave below, which sounds odd to my ears. If we're listening to, say, 1.5.6.4 we shouldn't drop down an octave at the 6.
I have no idea how you generate the MIDI file, but let's use the piano keyboard to describe what I'm hearing. When I play the WAV files from your progressions and we get to, say, an Amin chord, the Amin sounds like it is lower in pitch than the the Cmaj chord. This is OK, but it means that the 6 chord sounds an octave lower than I'm expecting it to.
My ears aren't trained enough to know if what I'm hearing is an inversion where you're playing the A note in Amin from the same octave as Cmaj but you're also playing the C and E notes from the same octave, or the A note itself is lower in pitch than the C. I would have expected to hear an Amin chord where the A note was a whole step above the G in Cmaj, and the C and E both higher than the A note.
Wow, great job on this for being someone who dosn't play an instrument. Seriously. Sure, there are some kinda not exactly right things, but sheesh, maybe you should make your own instrument! This is a great way to teach people about machine learning in a friendly way. Keep adding to it. Obviuosly you know there is random music- you were just speaking of songs. And You seem to know about blues, so you probably also know about rhythm changes and stuff? If not- check that out- see also Donna Lee.
"Musicians never play notes one at a time. They always use chords."
Today I learned playing literally any brass instrument means I'm not actually a musician. Who knew that my knowledge of the trombone ain't actually musicianship? ;)
Harmony in modern music has become homogenous and it's getting worse. I would think the best use case for this is to feed it modern chord progressions in order to inspire musicians (or machines) to avoid using these same harmonic tropes.
When you realize how the current "mapping" of piano keys makes everything confusing and how it historically developed, it looks like musician version of "JavaScript", i.e. a really crappy standard, where you can do anything but in a really unpleasant way. Map your scales directly onto launchpad/push/circuit, and you never get a wrong note. Moreover, the equal temperament makes all sounds slightly disharmonic, so even a professionally tuned concert grand piano sounds rather bad and unauthentic, different to what original classical composers intended (they used harmonic tuning where transposing by one key changed harmonics completely unlike with prevailing standard today). If you do acid trance, it's OK though.
> Map your scales directly onto launchpad/push/circuit, and you never get a wrong note
If you mean "map your notes so that you can only play the 7 diatonic notes of your major scale", well, then sure it's impossible to play "wrong notes", but it's also impossible to modulate to different keys, to play secondary dominants, any modal mixture, to add chromaticism, etc.
Similarly, as I understand it one of the advantages of equal temperament is that it allows you to modulate to any key within the same song. Of course if you never modulate you should just use just intonation in your key.
With equal temperament you actually don't modulate precisely, so there is always weird disharmonic sound every single time.
As I said, for some electronic music subgenres it's a welcome feature; it would shock classical composers if they ever heard such a version though. Have you ever thought why do you need to state scale at the beginning of a classical piece if you can just easily transpose song to whatever scale you like? Well, those scales sounded differently and each composer was exploring what sounds best in a given scale.
> it would shock classical composers if they ever heard such a version though
I am not sure what you are saying. What tuning system are you claiming composers like Bach, Mozart, Chopin and Shostakovich wrote keyboard music for, that playing their works in modern 12-tone equal temperament would "shock" them?
"There is a misconception that Bach supported equal temperament because of his famous 1722 collection, Das Wohltemperierte Klavier (The Well Tempered Clavier). Bach's clavichord was not equally tempered, but well tempered, like the title of the work says. The 48 pieces, two in each major and minor key, were written to show the character of each key in this temperament, the effect being completely lost in equal temperament. One can experience this by tuning a keyboard to well temperament, and then transposing Bach's Prelude and Fugue in C-major to C#-major, and vice-versa. The results will clearly show what the master was up to, and that the pieces were written for the nuances of each key with its particular coloring. Bach's theoretical framework was for pure intonation, and his music was written to be played in pure intonation, either by altering the tuning of the keyboard for each piece, or by using a flexible temperament that allowed pure tones in most popular keys."
"Equal temperament, a tuning which disfigures the natural intervals of the harmonic series, was invented as a necessity to the mechanical and engineering limitations of Middle Age instrument builders. But now we are in the 22nd Century and we can accurately measure the 1000th part of a millimeter. So why don't we use this technology to create in-tune musical instruments?"
I strongly recommend to read that discussion, it has both pros and cons of equal temperament and people are discussing the issue from both sides.
Equal temperament is very recent. Medieval through Baroque used a system of temperaments called meantone that began with C major being in tune, and then distorted tones slightly in order to make the proximal keys usable. Choosing a key in meantone implies a real difference in what the relative distance among notes was. When you go out beyond three sharps, it becomes very distorted. Five sharps is essentially unusable in meantone systems.
Bach's well tempering was a replacement for meantone where all the keys were usable. We don't know exactly what it was, though there have been some interesting attempts at reconstructing it. I remember being in the orchestra for the Bach four harpsichord concerto as a demonstration of one such attempt, which was wild.
As for using technology to create in-tune musical instruments, the original one, the voice, already has that. For keyboard instruments, there have been all kinds of crazy ones built over the centuries that have many more notes to an octave to allow much more precise intonation. But that's not the core restriction. I mentioned elsewhere in this thread that the pitch a given note maps to changes during a piece of music if you don't have tempered instruments involved, and those changes are always compromises.
So I wouldn't suggest reading that discussion, as it's apparently full of inaccurate material.
Bach wrote for meantone and later well tempered. Mozart was working with further variations on meantone, as was Chopin. Shostakovitch worked in equal temperament.
Mostly good info in there, but I see a few problems:
> There is also no E#. It just doesn't exist. Get used to it. Those are the 12 notes.
Technically not true. There is an E#, it's called F. If you are playing in F# major[1], the scale has an E# in it.
> Musicians never play notes one at a time. They always use chords.
Most lead guitar parts are not chords. Typically playing in a band context will create a chord from multiple instruments, but that isn't always the case.
[1] https://en.wikipedia.org/wiki/F-sharp_major
> Musicians never play notes one at a time. They always use chords.
This guy has also obviously never met the strange and rare creature known as a 'bass player'.
Also violin/string players (when not playing double stops)
Singers who are not overtone singers (pretty much everyone).
What kind of weird musician is this person limited to?
If you want to get a bit more wild, though, a lot of solo music still contains chords by implication, and can be very easily harmonically analyzed. I don't know how popular solo works are in the 20th century since atonality flung the music locomotive gleefully through the "Bridge Out Ahead" sign and into the canyon [1], but if we discount that it would certainly include almost all Western solo music.
[1]: Yes, it's an opinion. Yes, it's an informed opinion, and implying in one way or another that I just must Not Get It will not faze me. I do Get It, and that's why I find it distasteful.
I used to think that, but when I had to study atonal music, it became clear to me how much of an impact it had on film music and how it shows up in fragments in pop music.
Then, I took a class on electronic music composition. For my own project in the spirit of the avant garde, I wanted to challenge myself to write something "musical" without resorting to the traditional concepts of pitch-based melody and harmony or quantized rhythm. It turns out, there are many other levers you can play with to create the effect of an musical arc. Not to pat myself on the back, but I think the end result is evocative in a similar sense as tonal music. If I had the time, I would do more.
Check it out here, if you like: https://soundcloud.com/acjohnson/wanderers
If we want to play "wave the credentials" at each other, I minored in music composition, and my objection is a great deal philosophically deeper than "I don't generally like how atonal music sounds". There's a parallel world where atonal music is still explored but I wouldn't characterize it as "flinging music off a cliff". Atonality was merely when the cliff-jumping occurred.
I have neither the time nor the desire to fully explain my point in an HN post, but I would note that the music community itself has been asking questions in this direction in the last decade or so, if you want to follow up on it, as the academic music community asks themselves exactly how they got to where they are now, so disconnected from the rest of the world at large and so utterly irrelevant to almost anything musically occurring today. (Not just pop music; that is structurally inevitably insipid, really, so who cares, but irrelevant to anything other than their own very little world.)
also lo-fi sound chip music, but chord progressions may be emulated with fast arpeggiato (sequential instead of parallel chord notes, the name is from the harp, which exhibits this effect). There's probably a limit at which slow speed we don't notice notes as connected, but violin is surely fast enough. Speach pretty much seems to rely on an aliasing effect like that, too, when frequency clusters (iirc) are enough to identify words in speech analysis. So order would not matter as much as chord, if I am not mistaken. But then I might be as things get more complex than single progressions. I wonder if interleaved arp progressions have been explored.
lol, also every wind and brass instrument. Also drums.
There is a technique where the brass player hums while playing. In this way an extremely skilled brass musician can accompany herself.
Oh! Thank you; I hadn't realised that.
I learned this by asking a friend in the CouchSurfing meetup who brings a guitar or a roll-out piano. And as I say, I can't play any instrument myself.
Maybe if I want to learn, I should choose an instrument that uses single notes, it might be easier.
Regarding singing though, my ex-girlfriend told me to sing more quietly in church because it sounds so bad.
> Maybe if I want to learn, I should choose an instrument that uses single notes, it might be easier.
It's not just limited to bass/lead guitarists either - finger style acoustic guitarists don't play just chords. OK, often the fingerings are chord patterns, or arpeggiated chords, but that's not always the case. Checkout the title track from Bensusan's album, "Intuite", that is mostly just a single note melody: https://www.youtube.com/watch?v=di1qQsGcaIc (it would appear the ending is cut off this track here, but it gives you the idea).
That's nuts. That's like asking a beginning programmer to stop coding because their code is garbage. The only way to be a great singer is to be a bad singer first.
> I should choose an instrument that uses single notes, it might be easier.
I'd say the easiest / most immediately rewarding way is to pick up a guitar or similar (banjo / uke / banjolele / mandolin etc) and learn a few chords. Although people who can do that aren't exactly in demand so there's a case for learning the onde martineau or glass harmonica or something - that way if you get decent at it you can actually get paid to do it.
Or nearly all wind instruments....
Or any instrument that isn't tuned to 12-tone equal temperament. In just intonation there are no enharmonic notes, for just one example. C# isn't always the same thing as Db!
And that's without mentioning oddities like most varieties of bagpipes or the hurdy-gurdy, where a constant drone is present. It doesn't exactly form chords with the other notes, but it's still more than one note being played at a time!
Well, instruments in just intonation or other tuning systems can surely still play one note at a time!
Bagpipes and Hurdy Gurdy physically can't, the drone is always there.
Also, most wind instruments (that are played with the mouth -- organs excluded) are monophonic: trumpets, clarinets, saxophones, oboes, flutes, etc.
Or any solo player including a singer. It's a ridiculous statement.
Also know as the guy with the long guitar.
Source: I am a bass player
Or most instruments played in a symphony orchestra.
> There is also no E#. It just doesn't exist. Get used to it. Those are the 12 notes.
Whoa... Plus, they're identical only in equal temprament
> Musicians never play notes one at a time. They always use chords.
What kind of weird statement even is that? What about the first few notes of Fur Elise?
There are several statements you could pick out like that in this article. Like
> There is a chord called "C major", but it's totally different to the note called "C".
... the E and G of the C major chord are prominent overtones of the note.
> ... the E and G of the C major chord are prominent overtones of the note.
depends on the instrument. A pure sine wave at the frequency of C does not have any overtone.
That’s a fair point. But still, the relationship between the root, 5th, and major third are very much connected to the fact that they are specific multiples of that root frequency, and are the loudest overtones typically audible when the note is played. The root is not totally different from chords built on it.
I meant to add more to my top comment. I think the author does fairly well at their mission despite potential nit picks like that. It is really cool to see somebody learning and being excited about putting all the pieces together. The beautiful thing about music is you can go deeper and deeper as much as you want, and there are various competing ideas about what it is and how to talk about. Programming has a similar kind of depth to it.
plus one to this!
there are no pure sine waves without resonance, though
Or wind instruments for that matter.
And if you're going to say there's no E# you'd want to also pretend there's no B#. Otherwise it sounds like E is totally exceptional.
Down with E exceptionalism.
Manifest destin-E.
Yeah. Some of the stuff said in this is stated very poorly.
> A machine learning program is also called a "neural net".
A neural net is just one example of a machine learning model.
It goes along with this line... "For some reason I don't totally understand, some have 2 names (e.g. C#=Db, D#=Eb, F#=Gb, A#=Bb)."
It would probably blow the authors mind to know that some scales have sharps and flats, including the ones
I get the feeling this mistake was out of ignorance...not accidental.
sorry
Yes, the mistake was out of ignorance.
It was not an "accidental"
https://en.wikipedia.org/wiki/Accidental_(music)
And he didn't mention the obvious corollary... B# (aka C). (and Cb, Fb)
The dude's mind is going to be blown when he learns about double sharps and flats.
I got really confused by that at first, because I saw Abb in a note table.
When I looked it up, I saw that it's just another name for G.
https://www.basicmusictheory.com/a-double-flat-note
The fact that the same thing has 2 names, but then the same name (e.g. C) can mean 3 different things (note, chord, key) really confuses me. I'm a beginner (can't play any instrument), so this is just about trying to cover the basics and generate a machine learning dataset. If I weren't so confused already, I might try to learn an instrument. It seems to be really time-consuming to practice though, and I'd need a teacher to listen and correct my mistakes. I wouldn't notice when I get it wrong if I try to teach myself.
I'd suggest starting with the tin whistle. They're very cheap, very easy (you can learn the basics in a day or two, a week at most to get your first tune down), and any Irish session will either have someone willing to teach or someone who knows where you can get a teacher. The Session[1] is an excellent resource for it.
As for which whistle to buy, get one in the key of D. A Walton's little black should be about $6, one of their slightly better brass whistles around $11, and a Generation nickel D is about $12. If you really want to splurge Susato makes a very nice high D for $50 (their Kildare S series), but it's quite loud and more for if you are playing before an audience. The most expensive whistles around are probably Burke's, at $230 for a high D. So even if you end up collecting Burke whistles obsessively and getting one in every key he makes you'll still be out less money than buying some high-end guitars!
The one disadvantage is that they're diatonic instruments. That means you can play in one key, and only one key, without cross-fingering the whistle.
For tuning, I recommend either an electronic tuner or the program Flutini[2], assuming you get a tunable whistle (the cheap ones aren't and it probably won't matter much).
[1] https://thesession.org/ [2] http://www.novasession.org/Flutini/
Non-musician who taught himself all this stuff here, I feel your pain.
That said, Abb isn't another name for G - it's a distinct thing which, when played on typical musical instruments, is played the same way as a G. Basically note names to note frequencies is a many-to-one mapping.
If you want to understand the hairy details start by googling just vs equal temperament. But I went down that rabbit hole, and it hasn't made my procedural music any better, so I wouldn't really recommend it.
It's not a fixed mapping. It's more like a function that produces a mapping. It depends on the selection of:
- A pitch standard, like A440, which anchors one arbitrarily chosen note to a specific frequency.
- A tuning system, which defines the frequency ratios between notes.
Almost all modern music assumes _A440_ and _12-tone equal temperament_. In 12TET, adjacent flats and sharps map to the same frequency. It's said that they are _enharmonic_. But there are many other systems where this isn't true. In _just intonation_, notes are defined by simple frequency ratios, and there is no enharmonicity at all. Every number of sharps and flats map to different frequencies.
Even this is not quite right. For keyboard instruments, that have a fixed set of pitches that notes map to, you have to choose a temperament, a fixed set of pitches for each note (unless you're on a clavichord, where you can adjust it a bit by how hard you strike the key). Fretted instruments may or may not be tempered. Guitars are, since the frets on it limit the pitch range of a given note, though it is still a range. Viols aren't, because the frets are soft gut and used to brighten the sound, so the pitch range is much wider for a given note.
For string or wind instruments playing without keyboards, there is no temperament. Each note represents a range of possible pitches, and what pitch you choose depends on the notes around it. For example, if you are repeating or holding a note, you don't want it to change pitch. But if the harmony around that note is changing, it may be in tune at one point in time, then out of tune at a later one. String quartet playing involves a great deal of working out compromises about what pitch each note should have.
Good point! I admittedly was writing this off as "performance details" of music written in a modern 12-tone paradigm, but you're definitely right that plenty of music is written specifically for ensembles with continuous pitch with those nuances in mind. See also, a cappella vocal music, particularly barbershop. Even for guitar, some players tune to favor particular chords shapes sounding maximally consonant, even though the frets are designed for 12-TET.
its also about chord spellings within a context- how a musician, especially in a string quartet, thinks about how to play the note based on how it is spelled (eg is it the third, fifth, seventh or some extension of a chord?) the spelling tips you off to its context based on the key and other factors you feel but don't have time to contemplate when sight-reading.
Something I think a lot of people don't realize when starting out in music: you don't actually have to know how to play an instrument to make music. You don't even need music theory.
Knowing a little about how instruments work and learning some music theory can help tremendously, but you can actually just start poking around in a MIDI roll or tracker and go a long way.
Unless you're really into a specific instrument, starting off with any modestly priced keyboard with pressure sensitivity and MIDI out (USB-based if it's modern) will let you have fun with synths and sampled instruments you'd probably never be able to afford.
If you really want to learn a physical instrument that's easy to pick up, you can get a $20 plastic ocarina and find lots of tutorials on YouTube.
Just try playing any instrument - you'll soon realize that we're quite good at figuring out what sounds good and what not.
There are plenty of not-quite-correct things in this article, but I think both lines you quoted were meant to be more or less tongue-in-cheek.
Thank you. That's exactly right.
Indeed, those lines are dramatically oversimplifying, but it makes the comments section more exciting and the article shorter.
Perhaps it'll scare off people who were planning to do machine learning audio analysis. But at least they'll get an idea of how complicated it can be.
actually, there is an E#. It's called E#! There is also a Cflat! F is the enharmonic spelling of E# and B is the enharmonic spelling of Cflat.
If you want to try these chord progressions in your browser, check out:
http://wildsparx.com/rhythatom/
Coincidentally I just finished it. Source is here:
https://github.com/wildsparx/rhythatom
In theory it should run in any Chrome/Chromium browser, but I've received several reports of rhythatom failing to play. Would appreciate any help or ideas.
If you look at: https://peterburk.github.io/chordProgressions/ChordProgressi...
There doesn't seem to be any indication of major/minor. One way to interpret that is "always use the diatonic." Which means only use notes in the key, which means chords (1,4,5) are major while (2,3,6) are minor. However songs can have non-diatonic chords.
If you look at Rhyathatom, it defaults to the well-worn 1645 progression - the 6 is explicitly minor, which makes it diatonic. Try making the 6 major and you get a different animal - kind of sinister! That's a non-diatonic chord.
Maybe the author accounted for this elsewhere.
Rhytmatom is really cool! If I knew about it a month ago, I'd probably have used your source code instead of rewriting everything myself.
I like that I can choose the number of beats, unlike Jake Albaugh's arpeggiator (although that has some other interesting features).
https://codepen.io/jakealbaugh/full/qNrZyw
Nice work! Am I crazy or are the pulldowns for accidentals not doing anything?
> There doesn't seem to be any indication of major/minor.
There's a line in there about capital/lowercase of roman numerals not being important, by which I hope the author meant out of scope of the article. So I think they were explicitly just enumerating the chords of the diatonic scale.
Good catch. Will fix the accidentals.
I got this error the first time but worked fine on subsequent refreshes:
Thanks for the report; don't know why that happened; no iframes involved.
Sweet! I like that it only plays if the tab is active.
A for effort, but as a programmer and musician, simultaneously makes me a little sad as an example of the all-too-frequent phenomenon of programmers trying to model a domain without spending enough time actually learning the nuances of the domain first.
Nuances? Try basics. Worse, the author makes a number of general statements about music that I would have known to be wrong as a 6-year old with 1 year of experience in playing piano.
Who am I kidding? This is the age of crapping some mangled mis-information out onto a blog or web page in the hope of raising one's profile and securing a job with the next hot startup creating $10k robots delivering $5 pizzas or whatever.
The author clearly and good-humoredly declares that he comes with no knowledge of music or music theory at all, and is clearly documenting his process of coming to grips with enough of those basics you may have known as a 6 year old with 1 year of experience, and is attempting to do so in order to do something interesting. So perhaps 'wrongness' is far less important to this than 'interestingness' and, though I am a multi-instrumentalist myself and know music well, I'd like to thank him for attempting to learn what he needed to to attempt this interesting thing. I really enjoyed the read and intend to have a little play with the data.
Thank you so much for this encouragement.
The whole music investigation was a side-project, compared to my main efforts on Pingtype (a Chinese learning site). https://pingtype.github.io
I got interested because my friends at the KHOP prayer room gave me the song sheets in PDF format. They have the chords written, and the key on the top right. I thought it would be interesting to figure out which songs go well together so I could make some playlists. Even though I can't play music, I listen a LOT - all day in the office, and while I bike to work.
The fact this side project got more upvotes than my language-learning interests, and your encouragement that the data might be useful, is a sign to me that I should keep on with this and maybe learn to play. There's a piano in the house I rent, but I don't know what to do with it. It seems like now might be a good time to start learning.
Ok, perhaps you are fine with letting little gems like "Musicians never play notes one at a time. They always use chords." slide, but I'm not. If I can't trust the author about the basics and things that I know about, how can I trust their reasoning or thoughts on more complicated matters and items that are outside my own domain? I cannot, and hence don't read further because there is an opportunity cost in doing so, and there are many other interesting things I could be reading or otherwise doing with my time.
Honestly I just feel like you may have wielded judgement too hastily at the expense of your sense of wonder and shared joy in someone exploring territory that is well trodden only to relatively few. Very few people have a solid grasp of musical theory. For most people, scaling what may to you be barely the foothills of knowledge is a great and challenging experience, and I think it should be celebrated.
And sure, perhaps you and I can easily distinguish the subtle difference between 'Musicians always use chords' (obviously wrong) and 'Musicians always work, individually or together, within a key, which is an abstracted but well-defined set of tones that can be combined in time series or simultaneously to produce an inexplicably pleasant experience in the listener, by some neurological magic that has never been adequately explained' (non-obviously correct) but y'know what, I actually am willing to let little gems like that slide because they're adequate approximations of the truth that allow good folks to proceed on a journey that may one day lead them to, just maybe, have the arrogance required to belittle the uninitiated.
For me, the issue was that the writing style, like most articles written by programmers, passed off every thing as fact. Things like "they have the same chord progression" in section 2.a is patently false. This kind of writing is misleading to others, especially newcomers (blind leading the blind)
I do agree though, that to do somethimg, learn something, have something almost correct is better than having learned and done nothing, but as Bill Evans points out in the youtube video about the learning process, it's much better to start with something simple, concrete, and which you know as true then build from there, than start approximating (incorrectly) lots of things at once. So it's a case by case basis, but it's not merely gratuitous arrogance that motivates people to call other people out on false statements.
> Very few people have a solid grasp of musical theory.
This is just like saying that very few people in the world have a solid grasp of computer programming... Do you know that there are hundreds of university departments in the world where the only thing people do is, wait for it... having a solid grasp of musical theory!?
How is that not 'very few people' in a sample that includes the population of the Earth?
Not everyone if the world is trying to create a computer model of music. The minimum a person with this interest can do is to contact a specialist in music theory (of whom there are thousands as I mentioned).
Someone did a side-project in a new domain without studying it for a few years. Outrageous.
Good enough.
Also - the notation in question is confusing and overcomplicated, and ignoring corner cases makes a lot of sense if you don't care about outliers.
It's not like that notation handles all kinds of music well anyway.
Yes, exactly. The well temperated musical tuning [0] is "good enough". it's an approximation. Some pitches for notes sound almost the same and that's why they today are represented with the same key on a piano.
> For some reason I don't totally understand, some have 2 names (e.g. C#=Db, D#=Eb, F#=Gb, A#=Bb).
[0]: https://en.wikipedia.org/wiki/Well_temperament
Some famous music for the well temperament: https://en.wikipedia.org/wiki/The_Well-Tempered_Clavier
Yeah, my thoughts exactly.
> A chord is 3 notes played at the same time.
It's not like you have to dig too deep to discover four-note 7th chords, diminished/augmented, inversions, etc...
In the article:
"There's also 7th chords, Augmented chords (+), and Diminished (º) chords. As far as we're concerned, there's a lookup table that lets us turn these chords into notes."
This is nice as a diary of some hacking, but much of the musical commentary is misleading. Basically the author has enumerated all the unique ways of selecting the numbers 1..7 four times with replacement, and then done various mechanical translations on the results. Ostensibly this is done to enumerate all possible four-bar chord progressions, though in practice the results are highly redundant (i.e. many results are musically identical) and incomplete (many possible chords aren't included - e.g. secondary chords).
Then rather than distributing the scripts the author presumably used to do the enumerating and translating, they are distributing the outputs of the scripts, as a huge tarball full of midi files. Not sure what the motivation here was.
The script is in there! Chord Progressions.scpt.
If you can read AppleScript, go ahead and use it.
The way I see it, most people prefer to take existing data (e.g. Kaggle datasets) and build machine learning tools on that, instead of having to compile everything from source and make their own data.
If you want me to port the code to a more popular programming language, please tell me which one.
Rust? JavaScript?
C#? ;-)
In the Parsons Code section, if I'm reading correctly, the chords are conceived along an up-down axis. However chords do not really have that relation. Going from 1maj to 5maj can be up a fifth or down a fourth.
To illustrate this, first listen to the default 1645 progression on rhythatom. Then change the octave of row 2 (4maj chord) from +0 to +1. Can you hear its similar role in the progression, even though it's now stressed by being up an octave?
I talked about this with my friend Lenard, because I was confused why the Parsons code would be different for the same chord progression.
He then explained that I should move the key, and when I transposed the MIDI table, the Parsons codes all became the same.
I then asked him, is it possible to move up an octave for just one part of a chord progression? He said yes, it's possible, but people don't do that often. Normally people wait until the end of a chord progression to change octave.
Talking about going up or down an octave here is just nomenclature. Musically speaking, changes in a chord progression don't go up or down - this is because roman numerals refer to pitch classes (like C, E, G), not to specific pitches (like C4, E5, G3).
The matter of which notes to play for a given chord is called "voicing", and it's orthogonal to anything mentioned in your article.
Another correction: > The notes in "Cm" are C,D#,G.
It's actually C Eb G. The difference is that D# is an augmented 2 while Eb is a minor 3. They sound the same but they have totally different functions.
Remember there are different scales, too. Those 12 notes aren't the only notes.
https://en.wikipedia.org/wiki/Scale_(music)
Everything can be brought back to MIDI though, right?
Yes, but only with pitch bends. For example the Great Highland Bagpipe is tuned to a just intonation scale, which commonly approximates A Mixolydian (neither a major nor minor scale). (It's actually 3 pentatonic scales overlaid, but let's not get into that...)
Some info: http://publish.uwo.ca/~emacphe3/pipes/acoustics/pipescale.ht...
Getting that to sound right in MIDI requires the use of pitch bends, since MIDI only understands 12 tone equal temperament.
I would invite the OP to read a basic book on music theory, any good undergrad-level text should do, before stating some of the "facts" he is making, which are just untrue.
For one thing, the author is confusing physical pitches with musical notes. C double-sharp exists and is enharmonic to D, for example, because the frequencies are the same. But these are very different notes in a musical context.
There are many examples of this in the article that would be cleared up by a basic understanding of music theory that all music students complete, many of whom do this in piano lessons growing up long before graduating from highschool.
I admire the effort, but I often see people analyze music from a perspective that is not actually that of a musician, and there are always huge gaps in the understanding that needn't be there if they sought a basic knowledge of the matter.
Example:
> For some reason I don't totally understand, some have 2 names (e.g. C#=Db, D#=Eb, F#=Gb, A#=Bb).
Fair enough, you don't understand this, so that should be a clue that more knowledge is necessary before making this claim:
> There is also no E#. It just doesn't exist. Get used to it. Those are the 12 notes.
Incidentally, it goes much further than this, and for good reasons. F is the same pitch as E# and G-double-flat, but all three are distinct musical notes which serve different purposes.
There's also C-flat, B, A-double-sharp.
And B#, C, D-double-flat, etc.
Putting aside the rest of the nits here, there are two things that stick out in my mind. First of all, what he calls 4-chord songs are really more "4-bar songs" (which given his restriction of one chord per bar end up being at most 4-chord songs), but we should be clear about the distinction between the tones being played and the duration of those tones.
Second, I don't know if this is an artifact of my fluidsynth setup or not, but when I translate his midi files into WAV files the 6 and 7 chords are from the octave below, which sounds odd to my ears. If we're listening to, say, 1.5.6.4 we shouldn't drop down an octave at the 6.
You're actually using the data! That's really encouraging.
It's probably a bug in my script. How do I know which octave to choose?
If you can explain it to me using the Parsons Code method, I think it's easier for me to understand. 1.5.6.4 is udud, right?
I have no idea how you generate the MIDI file, but let's use the piano keyboard to describe what I'm hearing. When I play the WAV files from your progressions and we get to, say, an Amin chord, the Amin sounds like it is lower in pitch than the the Cmaj chord. This is OK, but it means that the 6 chord sounds an octave lower than I'm expecting it to.
My ears aren't trained enough to know if what I'm hearing is an inversion where you're playing the A note in Amin from the same octave as Cmaj but you're also playing the C and E notes from the same octave, or the A note itself is lower in pitch than the C. I would have expected to hear an Amin chord where the A note was a whole step above the G in Cmaj, and the C and E both higher than the A note.
Wow, great job on this for being someone who dosn't play an instrument. Seriously. Sure, there are some kinda not exactly right things, but sheesh, maybe you should make your own instrument! This is a great way to teach people about machine learning in a friendly way. Keep adding to it. Obviuosly you know there is random music- you were just speaking of songs. And You seem to know about blues, so you probably also know about rhythm changes and stuff? If not- check that out- see also Donna Lee.
“There are 30 keys”
You could make the case that there are 12 keys, or some multiple of 12 keys, or near-infinite keys. But 30? Pretty random!
Are some of those keys duplicates of each other? Or missing?
I could be wrong - I just guessed that from another data source (some PDF song sheets that have the key listed on the top right).
I think the 30 keys comes from the rollout piano from his couch surfing bud.
> That means every song in the world sounds like one of those 81 codes!
Yeah, every 4-chord repeating song.
"Musicians never play notes one at a time. They always use chords."
Today I learned playing literally any brass instrument means I'm not actually a musician. Who knew that my knowledge of the trombone ain't actually musicianship? ;)
Otherwise a great article.
Harmony in modern music has become homogenous and it's getting worse. I would think the best use case for this is to feed it modern chord progressions in order to inspire musicians (or machines) to avoid using these same harmonic tropes.
Anyone remembers Pachelbel Rant? First time I got aware of this topic. https://www.youtube.com/watch?v=JdxkVQy7QLM
When you realize how the current "mapping" of piano keys makes everything confusing and how it historically developed, it looks like musician version of "JavaScript", i.e. a really crappy standard, where you can do anything but in a really unpleasant way. Map your scales directly onto launchpad/push/circuit, and you never get a wrong note. Moreover, the equal temperament makes all sounds slightly disharmonic, so even a professionally tuned concert grand piano sounds rather bad and unauthentic, different to what original classical composers intended (they used harmonic tuning where transposing by one key changed harmonics completely unlike with prevailing standard today). If you do acid trance, it's OK though.
> Map your scales directly onto launchpad/push/circuit, and you never get a wrong note
If you mean "map your notes so that you can only play the 7 diatonic notes of your major scale", well, then sure it's impossible to play "wrong notes", but it's also impossible to modulate to different keys, to play secondary dominants, any modal mixture, to add chromaticism, etc.
Similarly, as I understand it one of the advantages of equal temperament is that it allows you to modulate to any key within the same song. Of course if you never modulate you should just use just intonation in your key.
With equal temperament you actually don't modulate precisely, so there is always weird disharmonic sound every single time. As I said, for some electronic music subgenres it's a welcome feature; it would shock classical composers if they ever heard such a version though. Have you ever thought why do you need to state scale at the beginning of a classical piece if you can just easily transpose song to whatever scale you like? Well, those scales sounded differently and each composer was exploring what sounds best in a given scale.
Watch this:
https://www.youtube.com/watch?v=6NlI4No3s0M
Ever played on a top end concert grand and thinking it sounds weird? Then tried a different tuning and being shocked how much better it sounded?
> it would shock classical composers if they ever heard such a version though
I am not sure what you are saying. What tuning system are you claiming composers like Bach, Mozart, Chopin and Shostakovich wrote keyboard music for, that playing their works in modern 12-tone equal temperament would "shock" them?
From the discussion on the video link:
"There is a misconception that Bach supported equal temperament because of his famous 1722 collection, Das Wohltemperierte Klavier (The Well Tempered Clavier). Bach's clavichord was not equally tempered, but well tempered, like the title of the work says. The 48 pieces, two in each major and minor key, were written to show the character of each key in this temperament, the effect being completely lost in equal temperament. One can experience this by tuning a keyboard to well temperament, and then transposing Bach's Prelude and Fugue in C-major to C#-major, and vice-versa. The results will clearly show what the master was up to, and that the pieces were written for the nuances of each key with its particular coloring. Bach's theoretical framework was for pure intonation, and his music was written to be played in pure intonation, either by altering the tuning of the keyboard for each piece, or by using a flexible temperament that allowed pure tones in most popular keys."
"Equal temperament, a tuning which disfigures the natural intervals of the harmonic series, was invented as a necessity to the mechanical and engineering limitations of Middle Age instrument builders. But now we are in the 22nd Century and we can accurately measure the 1000th part of a millimeter. So why don't we use this technology to create in-tune musical instruments?"
I strongly recommend to read that discussion, it has both pros and cons of equal temperament and people are discussing the issue from both sides.
That's incorrect.
Equal temperament is very recent. Medieval through Baroque used a system of temperaments called meantone that began with C major being in tune, and then distorted tones slightly in order to make the proximal keys usable. Choosing a key in meantone implies a real difference in what the relative distance among notes was. When you go out beyond three sharps, it becomes very distorted. Five sharps is essentially unusable in meantone systems.
Bach's well tempering was a replacement for meantone where all the keys were usable. We don't know exactly what it was, though there have been some interesting attempts at reconstructing it. I remember being in the orchestra for the Bach four harpsichord concerto as a demonstration of one such attempt, which was wild.
As for using technology to create in-tune musical instruments, the original one, the voice, already has that. For keyboard instruments, there have been all kinds of crazy ones built over the centuries that have many more notes to an octave to allow much more precise intonation. But that's not the core restriction. I mentioned elsewhere in this thread that the pitch a given note maps to changes during a piece of music if you don't have tempered instruments involved, and those changes are always compromises.
So I wouldn't suggest reading that discussion, as it's apparently full of inaccurate material.
It doesn't answer my question. What tuning system did they write for?
Bach wrote for meantone and later well tempered. Mozart was working with further variations on meantone, as was Chopin. Shostakovitch worked in equal temperament.
Shostakovich is a 20th Century composer, died in the 1970s, so at least as far as he is concerned, he wrote music with today's tuning systems.