> If so, how come a church organ doesn't sound like a chip tune, which is also built up from simple waveforms? Well, actually it will, if you remove the church.
That's not true. In the real world an acoustic device will never generate a pure sound. An organ pipe will generate all sorts of artifact noises and its overall sonic behaviour will be affected by temperature, humidity, barometric pressure, and its physical promixity to other pipes (interference is a common problem in mixtures).
It is important to note that organ builders will graduate the different pipe dimensions within a single register in order to achieve a homogeneous sound from low to high notes, but that graduation does not necessarily follow physical laws but rather an approximation. Sometimes the graduation will involve jumps or repetition in order to create a register with a mixed character.
Finally, the voicing of individual pipes, done by hand, will (when done by a master voicer) balance character and homogeneity. Each register will have its own personality, but each note will sound slightly different.
> You have to add movement, energy, and emphasis (which, on an organ, has to be implemented by varying the duration of the notes, and the pauses between them, because there's no dynamic response).
On mechanical organs, a good player will also be able to adjust the opening and closing of individual pallets (valves) using their fingers, in order to control the attack and decay of individual notes.
I remember an accountant for a company that built and tuned organs talking about their search for field service guy. He said it's a difficult position to fill. A candidate needs perfect pitch. Familiarity with religious sects and church music because the organ needs tuning according to the music played and cultural expectations of the parish[1]. They need to be able to navigate church politics. And be willing to travel to the middle of nowhere on a regular basis. And mechanically adept.
[1] He said some music is designed to be played on a rich earthy sounding organ and others on one that more pure and ethereal. Tune the organ the wrong way and they'll horrified what you did.
Why? Traditionally, organs are tuned by listening to the beats[0] between pairs of notes. If you want to tune to specific frequency (such as the modern A=440Hz standard) you only need a single frequency reference, such as a tuning fork. But historically there have been an enormous variety of tunings used in pipe organs, many of which were built before A=440Hz became standard, and in any case, the tuning is affected by temperature, so it will drift as the seasons change.
I think that company is needlessly limiting their options by requiring perfect pitch. Good relative pitch perception is sufficient.
Too late to edit my post, I thought of a reason. I expect most pipes in old organs are unlabeled. Somebody with perfect pitch can identify them quickly and without needing any electronic device, leaving their hands free for other tasks.
An organ pipe, when played on low to moderate pressure and allowed to reach a steady state, generates a pure harmonic tone. There is only a single audible oscillator (either the column of air in the flue pipe, or the acoustically coupled system of the beating reed and the column of air in a reed pipe), and this oscillator is periodic, so all partials are exact integer multiple of the fundamental.
This is not true of all instruments. A string oscillates both by tension along the string and by bending across the string. The bending component becomes especially important with thick strings, so a bass guitar produces noticeably inharmonic sounds. This can make the sound more complex and interesting, but it can also result in "muddy" sounding chords. A hammered string, such as used in a piano, works the same way. Pianos traditionally compensate for this by using "stretched" tuning, i.e. going up an octave is slightly more than a doubling of a frequency. This helps the harmonics of the chords line up with the slightly sharp partials of the inharmonic tone.
Another way to avoid the inharmonicity is with bowed strings. Here the slip-stick motion of the bow resets the motion of the string every cycle, like hard sync in a chiptune. The bowed string in steady state is purely harmonic.
Tuned percussion instruments have even more oscillators interacting. A marimba bar twists and stretches along many different axis, resulting in a complex inharmonic tone. A skilled tuner can adjust the shape of the bar to bring many of the partials close to integer ratios. The same is true of bells.
A pipe organ can have a noisy "chiff" at the start of the note, but whether or not this is audible depends on how the pipes are voiced. It has been considered desirable or an unacceptable fault according to the fashions of the time, so there is nothing wrong with omitting it.
It is not practical to control how far the pallets are opened by partially depressing the keys. In a mechanical action organ it takes significant force to overcome the initial air pressure, which generally makes the pallet spring all the way open. Pneumatic and electric actions don't even theoretically allow partial opening. I am not aware of any organ composition which calls for partial opening.
IMO a filtered chiptune wave, with good quality reverb, makes a excellent organ-like tone.
Organ partials aren't perfectly periodic. There's some chaotic pressure variation around the lip, and the pipe also distorts mechanically with pressure variations along the various possible modes - much more than a stiffer instrument played with lower pressure, like a flute. So not only are the partials subtly inharmonic in various ways, they also have different attack envelopes at the start of a note, and the air pressure from the bellows/pump may also vary over time as notes/ranks are played.
This is one of those common situations in music where a trivial model is too superficial to capture the defining subtleties of a system. The richness of a full pipe organ is a combination of subtle effects within each pipe, combined with subtle variations within each rank, combined with the not-so-subtle effect of combining multiple ranks, combined with the acoustic environment of the hall, which doesn't just add a generic stereo reverb, but also supports standing waves at some of the lower bass notes.
A filtered chiptune wave really isn't a very good approximation.
Your post reminded me of this video[1], where the guy plays this giant double bass. You can clearly see how the low frequency oscillation of the string forms as he drags the bow across it.
This reminds me of a chapter "The Organ Makers' Shop" in Matthew Crawford's book _The World Beyond Your Head_. The author spent a lot of time with organ makers Taylor and Boody (Staunton, Virginia) in 2007 and 2008 and his descriptions are rich and interesting.
This is exquisitely interesting level of detail. I've always thought that there was more to the organ I hear than the conceptual mental model of a binary "on/off" smooth tone generated by the organ - thanks for articulating some of the nuances that go organ playing.
That said, I can't help but feel you're missing the point of the article though...
There's the old quote:
"All models are wrong, some are useful"
Of the traditional musical instruments we have, I think 8 bit waveforms probably lend themselves well to the organ (more so than other instruments: strings have decay and vibrato, even other keyboard instruments have decay and tonal qualities).
Additionally, contrapuntal baroque music probably worked quite well for the constrained environment the hardware provided by way of number of voices and length of music...
The most artificial sounding sections in this synth music are those were chords and harmonies are just too perfect and blend together too much. This is caused mainly by two things: perfect tuning and perfect oscillator stability. If one were to mess with both (slight random base offset for each note, even slighter phase drift per note) one would probably get rid of that and generate a more impure, "warmer" sound. The convolution reverb works wonders on the synth output, though.
Reader may be interested to hear how church organs from Bach's time sound. My home town has two instruments, which can play this old music, such that it may sound like back then:
There was quite a bit of change over the centuries. But they were among the first organs brought back to the baroque ideals and the original state as much as possible. There is a lot of information about these efforts, but mostly in German. Basically the Cosmae organ is now like it was in 1688. This was finished in 1975 and influenced a lot of other instruments to be brought back as close as possible to their original state.
This post reminded me of my Senior year of EE, we were supposed to do a senior project and I decided I wanted to make a custom IC to play music (our University has a chip fab facility). Keep in mind this was 1978 so the size of the chip was goin to be limited given our facility. I did not know a lot about music, I knew I liked music but I just didn't know a lot about the theory behind it. So I decided to take a music class as an elective. After I enrolled in the class I was told by fellow students that this class was "an easy A". I was tempted to drop the class because I was not looking for an easy A I was looking to learn about music. On the first day of the course the professor was an older, grey haired older gentleman. He explained that the course would be composed of some music theory and some listening. We were to go to listening rooms and listen to various pieces of classical music and become familiar enough with the music to be able to identify it when hearing it. Again, it was 1978 so the music rooms contained vinyl records and record players and headphones. Our first Quiz came and the professor played various records and we were to identify each piece. To my surprise and amazement the other students were able to easily identify each piece. I was not able to pick out even one. The music theory part I was picking up easily but the listening part I was struggling with. I spoke with the professor and explained to him what my project was and he seemed very interested. We got into a long discussion about it which I was really excited about that he was so interested in what I was doing. I explained to him that part of my dilemma was to fit the notes into a small amount of memory that I would be able to fit into the chip I was making. He suggested that I look into Bach, specifically Fugues because there was a lot of repetition, sometimes the same notes simply shifted in tone etc. That was excellent advice. I continued listening, more and more time in the listening rooms but for some reason I was just not picking up on how to identify the music as easily as the other students in the class. I finally asked a fellow student how he was able to do it so easily. He said that everyone in the class had notes that he showed me, the notes said things like "blue label with yellow writing", "red label, blue writing" and other such things relating to the appearance of the labels on the records. Again, it was 1978, and the professor would hold up the record to the light, then place it on the turn table before playing each piece, the other students would then refer to their notes which they had prepared in the listening room. They were cheating. I really liked this professor who I had forged a relationship with so I went to him and explained to him what was going on and that people had told me his class was an easy A. He thanked me and asked how my project was going and we talked about that. He didn't seem mad or perturbed particularly. He mentioned that he would appreciate if I brought my project by and show it to him when I finished it and I said I would. The final exam day came and I had continued the listening sessions and had decided not to cheat like everyone else, I really wanted to learn about music. I was really glad I made that decision because for the final there was a written theory portion and a listening and identifying part. The professor brought out a tape deck (reel to reel) and would start and stop it for each piece or portion of a piece we were to identify. You could hear the air go out of everyone's lungs in the class (except me). I got a little smile and proceeded to identify each piece to my own surprise. Later after I had presented my project to my EE professors (and got an A for it) I took it to my music professor and showed it to him, he really liked it, I had hard coded Toccata Fugue in D-minor. He revealed to me that I was the only one in the class that passed (I got a Not so easy but satisfying A). He flunked everyone but me ! I kinda felt bad about that for about one minute.
Thanks for the excellent story. Doing your own music player chip as an undergraduate in 1978 is no small achievement.
> He said that everyone in the class had notes that he showed me, the notes said things like "blue label with yellow writing", "red label, blue writing" and other such things relating to the appearance of the labels on the records
This is a classic mishap for machine learning systems as well, picking up an unintentional but clear "sideband" channel.
A classic mishap for human learning as well. I minored in music, at what wouldn't even be a particularly prestigious music school or anything. I am sure that had the music students not cheated in the first place, they'd still have found it easy to identify the pieces. Assuming the professor was drawing from the canon, a good chunk of them should already have known a number of them. But they took the easy path, for something that should already have been pretty easy....
(The theory is what they should have been having trouble with. The musicians, in general, hated theory class. It was exactly the sort of thing they'd generally tried to get away from by going into music. This is almost a quote from them. Exceptions exist, of course, but generally they were not enthused.)
I would love to experience him hook up a c64 to an amplifier in a church and play these songs. Would it sound the same as the recording on the website?
I'd love a "making of" follow-up. I would definitely be interested in a comparison of a straight-up composition without the movement, energy, or emphasis, and then adding to it in pieces to see the work come together.
That's not true. In the real world an acoustic device will never generate a pure sound. An organ pipe will generate all sorts of artifact noises and its overall sonic behaviour will be affected by temperature, humidity, barometric pressure, and its physical promixity to other pipes (interference is a common problem in mixtures).
It is important to note that organ builders will graduate the different pipe dimensions within a single register in order to achieve a homogeneous sound from low to high notes, but that graduation does not necessarily follow physical laws but rather an approximation. Sometimes the graduation will involve jumps or repetition in order to create a register with a mixed character.
Finally, the voicing of individual pipes, done by hand, will (when done by a master voicer) balance character and homogeneity. Each register will have its own personality, but each note will sound slightly different.
> You have to add movement, energy, and emphasis (which, on an organ, has to be implemented by varying the duration of the notes, and the pauses between them, because there's no dynamic response).
On mechanical organs, a good player will also be able to adjust the opening and closing of individual pallets (valves) using their fingers, in order to control the attack and decay of individual notes.