Could typically electronic sounds be generated mechanically?

Question

I was thinking about our history of music, came upon electronic music and started musing about all the possibilities of modulation and such.

I then thought back to a previous question of mine, "How might a purely mechanical 'voice' function?", and thought about Vocoders and similar synthesizers. Then I noticed that beat-boxers can create similar effects using a solely mechanical device and by it was motivated to write this question.

Would it be possible to have sounds found in modern electronic music generated mechanically? I'm looking for the following traits:

• Frequency adjustment
• Volume adjustment
• Synthesis, i.e. combining of multiple waveforms
• Filtering, i.e. volume adjustment according to specific patterns
• Flexible attack-decay-sustain-release
• Arpeggiation

A MIDI-like system needn't exist, i.e. existing sounds needn't be modulated during performances, they may all have been prepared prior to performance.

Answer 1

TL;DR : These parameters can indeed be controlled mechanically, though not to the extent enabled by today's synthesizers and software.

Pretty much all of the instruments of the traditional orchestra can adjust some or all of these parameters mechanically. They existed well before synthesizers were a thing, though the terminology was different.

Let's take the piano as an example: Frequency (pitch) is determined by the length and diameter of the string, and controlled by the keys. As for volume adjustment, the piano can play a range of dynamics, from quiet (pianissimo) to loud (fortissimo). That's actually the killer feature that made piano so popular during the Romantic period. Mechanically, this is caused by hammers hitting the strings with varying strength. Filtering occurs in conjunction with dynamics. The higher the volume, the brighter the sound (meaning more energy in the high part of the spectrum). The soft pedal can also be used as a "low-pass-ish" filter.

When it comes to synthesis in the mechanical world, the pipe organ can arguably be considered the very first synthesizer in Western music. By combining sets of pipes, it is possible to change the overall sound, much like performing additive synthesis.

Edit: As pointed out in the comments, ADSR envelope can be controlled mechanically to some extent, for example using the sustain pedal of the piano. The envelope of string instruments can be modified by using - or not using - a bow: think of the double bass in classical music (bowed) VS double bass in jazz (fingered).

It is worth noting that, in purely mechanical instruments (and arguably in analog synthesizers), it is almost impossible to adjust one parameter without another varying accordingly, for example frequency and volume, or volume and filtering. Think of an oboe player trying to produce the highest notes of the instrument at low volume, that's near impossible.

As for arpeggiator, you might want to take a look at punched cards and how they were used to generate melodies, typically played by barrel organs. This contributed to the start of electronic music at the beginning of the 20th century.

Answer 2

Not really.

You can reproduce most of the synthesis parts of electronic instruments through mechanical means, but you'd do this mostly individually, whereas synthesizers allow you to change many of the parameters at once. So, a piano gives you some control of the decay and release, and a kick drum lets you modify the attack by changing the beater, and a violin gives you far more control than ADSR. You can also filter sounds by throwing a blanket over the instrument, or closing a door to the room the instrument is in, but the player can't control this on a per-note basis as would be typical in a synth.

So if you want to reproduce the listed effects one at a time, you might find a mechanical method to do so. There are other sounds which are not so easily mechanically reproduced though. Changing the time on a delay, for example, has no mechanical equivalent, unless you move the performer or audience around at incredible speeds and g forces, to use doppler changes...

However, in a more general sense, the questions asks about typical electronic sounds. To me, sounds are more than synth parameters, or individual effects. Typical electronic sound to me would also includes aspects introduced during mix time, like sidechain compression. This is a very typical electronic sound, and I can't think of a way to reproduce it using mechanical means. Some other examples include noise gating, looping samples at non-zero crossing points, multiband compression, autotune on vocals, etc. These are typical sounds in electronic music, and are not possible to recreate mechanically, unless you go build a giant mechanical computer with cog wheels to do the maths. I dont think this weas in the spirit of the question :/

Answer 3

You could do this with fluidics. Fluidics allows you to carry out all signal amplification, processing, and even computation with fluid flowing through some funny looking pipes. In fact, you can build devices which operate analogously to electronics using fluidics. The image below describes a fluidic acoustic sound amplification system.

The device in the middle is one of the most basic devices in fluidics, the fluidic amplifier. The working principle is that a low pressure jet(control jet) of fluid can be used to deflect a high pressure jet(power jet) of fluid between two ports. They have been demonstrated to work in the KHz range. Operation above 30 KHz has been suggested by using small amplifiers using light gases such as hydrogen or helium.

We can accomplish volume adjustment with the system above by modulating the supplied pressure to the power jets. Frequency adjustment is also possible too because we can build pressure controlled oscillators. One way this can be done is by routing some of the output flow of our amplifier back to the control ports. In addition there are a number of frequency responsive fluidic elements and circuits. So it should be possible to make a fair amount of sound modulating devices using fluidics.

It shouldn't be too hard to make an attack-decay-sustain-release envelope generator from a couple of amplifiers and passive elements. NASA demonstrated a fluidic stepper motor which was capable of 300 steps per second, which requires applying pressure in a sequence to a series of bellows. The circuit used could be modified to drive a set of adjustable fluidic oscillators rather than bellows to make an arpeggiator.

Answer 4

The pre-arrangement time would be considerable but you should be able to get good results using something in the vein of the cylinder and comb device found in a common music box. Similar devices have been used to play elevator music and, through the use of interference, can even be used to recreate complex orchestral pieces to a reasonable degree of facsimile.

Answer 5

If I understand your question correctly; this should be possible in theory; it would just be extremely complicated to design a physical system to replicate it.

Your final paragraph says:

A MIDI-like system needn't exist, i.e. existing sounds needn't be modulated during performances, they may all have been prepared prior to performance.

I am interpreting the situation to be as follows:

1. You are given a sample or waveform of an initial sound which is to be modified.
2. A modification is requested, such as a change in pitch, change in volume, anything which we would typically do with electronics.
3. Something mechanical is constructed which can perfectly generate a sound which is the initial waveform with the required modifications.

As far as I'm aware, this is something that is completely possible to accomplish. Basically, if you can do electrical processes to modify the electrical waveform in some desired way, you can do the same thing to a mechanical (acoustic) waveform, through careful design parameters (the same way it is done electrically).

In essence, the main power of electromechanical analogies is that you can model electromechanical systems in a mathematical framework. This is essentially what allows us to use electronics to modify electrical signals which eventually are produced as acoustic signals after going through the transducer of the speaker.

Using electromechanical analogy, one could determine the exact input and system parameters required to design a fully mechanical system which, when given the appropriate mechanical input, creates the exact acoustical waveform that you desire. What that system would look like or be made of is far more complex, and outside the scope of what I could guess at.

There are some good tables on Wikipedia which actually show which variables in each system represent the analogous system in other domains. To actually come up with what the physical system would look like, you would require a lot of knowledge about acoustical dynamics; but in essence it would just be replicating the functions of the electrical components with acoustic components.

Answer 6

Yes, because sound is mechanical

Synthetizers process electric signals to construct electromagnetic waves, which are then sent to a speaker, which will vibrate a membrane to produce the sound.

As long as you can vibrate the speaker just as you would do with the synthetizer, you can mimic its sound.

All computations possible can be done on non-electrical computers (i.e.: a literal cog machine, if you're into that). It may just be that to mimic a microchip mechanically, including its speed, you may need a prohibitively expensive and immense setup. But that does not mean it is impossible.