I’m new to this and I’ve had fun building all the examples here and on YouTube. I wanted to try to synthesise real instruments and after reading an article in Sound on Sound I got an idea for a patch. Although I’ve had good results using Vult’s Basal and Slap, I wanted to build from first principles using basic modules.
This is the idea - each key on a grand piano hammers a set of 3 strings; a high, a medium and a low string tuned an octave apart. These are slightly out of phase with each other.
The 3 XCOs used here represent these 3 strings, each tuned to C an octave apart and each slightly out of phase with the others. I used the XCOs because they have the ability to mix square, sawtooth, triangle and sine waves which gives a rich and complex wave.
The 3 strings are mixed to different levels before being sent on to the main mixer. This allows for the low tones of the note to resonate more than the high tones.
This is far from perfect I know and the complexities of all the resonances present in a real grand piano are probably only theoretically possible to build this way. I will try to build in a mute/sustain pedal of some kind for the next version.
[Edit] This contains many errors and bad assumptions - thinking again about this one. I can’t see a way to do this without treating individual strings differently and then how do you trigger from a standard MIDI interface?
Dr. Manny Fernandez has an interesting series of videos in which he details how he constructed a grand piano emulation using a Yamaha Montage: the building blocks there being 8 independent layers of 8-operator FM synthesis, each layer having access to its own pair of insertion effects and separate send levels for the two system effects.
Even if you aren’t planning to follow his structure or use FM at all, it’s still interesting to see the different elements of a piano’s sound which he picks out for modelling. This includes the inharmonicity of the strings as well as what he affectionately calls “the stuff”: sounds such as the thunks of the hammers. This provides a lot of ideas for ways to expand on a simple piano emulation.
The total run time is ~70 minutes. There’s also an accompanying series of articles.
That’s completely incorrect about pianos. I wonder where you obtained that idea. The trichord set is tuned very precisely in unison to a high degree of accuracy. I speak as a former professional piano tuner and instrument maker. On a piano octaves are always true and beatless, but due to the departure from ideal infinitely thin strings of the relatively thick steel wire in pianos, tuners make the octaves ever so slightly sharp and increasingly so in the upper range, and likewise modify the bass, but it is a very small effect. Find a grand piano, and pluck the individual strings. You will see. And there are no exceptions to this, either now or historically. Harpsichords had registers of single strings at 8 foot, 4 foot, and sometimes 16 foot pitch. A piano is not a harpsichord.
If you have questions about how pianos actually work you are welcome to contact me.
Most pianos are sampled, as the synthesis of the complexity across the compass is overwhelming. There are very successful piano models achieved with Physical Modelling technique, such as Pianoteq. Nearly indistingishable from a sampled instrument. The Sound on Sound article is very simplistic. Nevertheless interesting.
Thank you very much for your insight. I have obviously made several mistaken assumptions here. Your advice will help me improve my modelling. I am very grateful for your expertise.
A little fiddling and I got a much better sound; Upped the triangle wave on the low string, maxed the sine on the high string, some saw on mid and low, no square on any strings. A tiny bit of attack helps with the hammer action, less on the high string, more on the low string.
The high notes (octave 5 and above) need to be softer and the low notes (octave1+2) need to resonate more - suggestions welcome.
It wouldn’t be simple anymore, but you could always do it the same way a real piano does: each note has its own set of hardware tuned in frequency (and in this case amplitude).
I am working on a version with 88 sets of 3 XCOs, but I need to stabilise this one first. It’s great for the middle range, but sometimes when I come back to it, it sounds different (a quarter tone sharp, even).
I had the best results for piano sounds with FM synthesis. My tip would be to take NYSTHI TZ-OP and modulate it with 3 more TZ-OP’s with Ratios 1, 1.5 and 3. Then bathe everything in reverb and you might get pretty close to a piano experience. And if you want to have it velocity sensitive modulating the FM offers some good options to make it dynamic.
Could it have been conflated with this business from Piano - Wikipedia ? …
All else being equal, longer pianos with longer strings have larger, richer sound and lower inharmonicity of the strings. Inharmonicity is the degree to which the frequencies of overtones (known as partials or harmonics) sound sharp relative to whole multiples of the fundamental frequency. This results from the piano’s considerable string stiffness; as a struck string decays its harmonics vibrate, not from their termination, but from a point very slightly toward the center (or more flexible part) of the string. The higher the partial, the further sharp it runs. Pianos with shorter and thicker string (i.e., small pianos with short string scales) have more inharmonicity. The greater the inharmonicity, the more the ear perceives it as harshness of tone.
The inharmonicity of piano strings requires that octaves be stretched, or tuned to a lower octave’s corresponding sharp overtone rather than to a theoretically correct octave. If octaves are not stretched, single octaves sound in tune, but double—and notably triple—octaves are unacceptably narrow. Stretching a small piano’s octaves to match its inherent inharmonicity level creates an imbalance among all the instrument’s intervallic relationships. In a concert grand, however, the octave “stretch” retains harmonic balance, even when aligning treble notes to a harmonic produced from three octaves below. This lets close and widespread octaves sound pure, and produces virtually beatless perfect fifths. This gives the concert grand a brilliant, singing and sustaining tone quality—one of the principal reasons that full-size grands are used in the concert hall. Smaller grands satisfy the space and cost needs of domestic use; as well, they are used in some small teaching studios and smaller performance venues.
Not that I believe that Wikipedia is the final authority on everything, but some version of that may account for “that idea”.
Even so, tuning real strings “slightly out of phase with each other” does not sound possible to me, unless the second and third are struck with very short delays after the first one by two more hammers. Out of tune, yes, but not out of phase.
