Elementally, My Dear...

Warning: Do not use image above in place of a periodic table.

Part of the craziness that was the end of last semester included my final projects for the electronic music class I was taking. One of these projects was the electronic dance piece that is getting premiered in March, about which I will write more after I have finished revising it. Another one of these projects was an original suite I composed called The Elements. Before all you chemistry nerds out there get super excited about a collection of 90 (or 100 or 110 or however many total elements there are out there now including the synthetic ones), let me warn you that there are only four movements: one for each of the natural elements, according to the ancient Greeks. Subsequently, the four movements of The Elements are (1) Air, (2) Fire, (3) Earth, and (4) Water.

Each of the four movements also fulfilled a particular requirement that was part of the project option I selected (the other components of the final portfolio were essentially free composition). Specifically, this project option had to be exactly four movements, each one having limitations in terms of the source material that could be used. As I go through each movement below, I will explain in greater detail how the sounds for each had to be constructed. (The movements were ordered according to my artistic preferences and not based on the order in which the restrictions were listed in the assignment.) And, after each movement’s explanation, I have made the music available for listening.

Air

The sounds that comprise this first movement are entirely “found audio,” that is, audio I personally collected myself, using a handheld recorder. This excludes any sounds downloaded or otherwise obtained from the internet or other preexisting collections of sounds. As this is the movement about air, the prevalent sounds here are recordings of me breathing and blowing over an open bottle. As has been the case with most of the electronic music I have composed, these sounds appear in the piece modified from their original forms. In the case of the air sounds I recorded, I added reverb and resonance, the latter allowing me to enhance any sounds with a particular frequency (think of it like having a note sound every time you made a sound like snapping your fingers).

Additionally, there are sounds that are not air-based so much as they are air-inspired. For instance, I modified a homemade recording of my door harp to suggest the sound of wind chimes blowing in the wind.  Likewise, the chorus at the end of the movement is part of a recording of a piece I wrote titled Nishmat, which is the Hebrew word for both breath and soul.

Fire

The sounds in this movement are derived exclusively from spectrum synthesis (which I suppose ought not to be confused with spectral analysis). In the case of MetaSynth, as explained in its reference manual, spectrum synthesis “uses high-resolution frequency analysis to construct a series of spectrum events by analyzing slices of a recorded sound.” Below is a screenshot from MetaSynth showing such an analysis:

Spectrum Synthesis in MetaSynth

The horizontal axis represents time and the vertical axis represents frequency. Brightness corresponds to amplitude, or loudness (brighter equals louder). The sound of this particular spectrum is the spoken word “fire.”  The bright bands that appear in roughly the first third of the spectrum represent the (long) ‘i’ sound in the word “fire.” In normal speech, the (sounding) vowels are specific collections of tones whereas consonants are unpitched. It’s comparable to percussion instruments that have definite pitches (piano, chimes, vibraphone, etc.) versus those that do not (snare drum, wood block, cymbals, etc.).

So, how is using spectrum synthesis any different from just using the original sound? Excellent question! You will note in the image above that there are a series of faint vertical gray lines – these divide the sample into the aforementioned “spectrum events” (in this sample there are 64 events). When the spectrum synthesis is played back, the sound is constant for the duration of each event. If this sample were 64 seconds long, each event would sound for one second, and during each of those seconds the sound would be the same. (There are options to fade in and out of each event, allowing the sounds to blend a little more.) Regardless, there is an inevitable loss of information – when you hear sounds in everyday life, you don’t hear them in short little chunks. Whereas initially there are essentially an infinite amount of samples, after the spectrum synthesis there are at most 64 samples (the program allows you to have as few as 2). This in turn is comparable to estimating the area under a curve (à la calculus): “normal” sounds are like the “true” estimate of the area under a curve, whereas the spectrum synthesis is analogous to an imperfect estimation using (for example) a finite number of rectangular bars (see below).

Approximating the area under a curve using rectangular bars.

The result of all of this is a sound that can closely resemble the original (and is usually recognizable), but is inherently rougher. In other words, there is an added level of artificiality. As in the first movement, once I had a collection of these spectrum-synthesized sounds, I was free to modify them as needed. But nowhere will you hear sounds in their original form, i.e., as they might be heard in the real world (outside of an electronic music concert or something similar).

The sounds used in this movement are (synthesized versions) of fires burning and society’s subsequent responses, including a fire truck siren and a fire alarm. In contrast to the serenity of the opening movement, “Fire” is rushed and driven.

Earth

The construction of the final two movements involved the use of the image synth room, which I wrote about in a previous post. Below is one of the images I used:

Image Synthesis in MetaSynth

The horizontal and vertical axes represent time and frequency, respectively, just like in the spectrum analysis example I displayed earlier. Likewise, brightness still corresponds to loudness. There are two major differences between spectrum synthesis and image synthesis, however. The first is that the colors represent different things. In spectrum synthesis, there are a limited set of colors one will see; all of them underscore the sounds brightness. In image synthesis, all colors (well, 16,777,216 colors to be precise) are possible. Using the RGB color system, red corresponds to the left channel, green corresponds to the right channel, and blue corresponds to the “grid channel,” which is not converted to sound but can be used to align different parts of the image.

The second major difference is that sounds produced using the image synthesis are played through an (electronic) instrument. This instrument can be purely synthetic, using simple types of sound waves and their basic combinations, or it can be slightly more complex, using samplers. The third movement corresponds to the former, and the fourth movement corresponds to the latter.

In this piece I have depicted Earth in the planetary sense: as the beautiful—yet fragile—blue orb floating in the vast expanse of outer space, glistening in the sun. While some of the images I used to make sounds were more abstract, I also include some Earth-themed pictures, such as the view of our planet from space and a beautiful landscape. This movement sounds the most like “computer music” due to the fact that I was limited to purely synthetic instruments.

Water

As mentioned above, this last movement was constructed solely from sounds obtained using sampler instruments. Think of a sampler instrument as an electronic keyboard with all the regular notes, but when you depress the keys, instead of a piano sound you hear any sampled sound, like a bird chirping or (as is heard in this movement) a toilet flushing.

(In fact, all the preset sounds on an electronic keyboard are essentially sampler instruments – the pre-recorded the sound of a particular piano note is just being replayed at different frequencies. Slightly more sophisticated sampler instruments—multisamplers—use multiple samples over different ranges. For example, a clarinet in its low register sounds markedly different from a clarinet in its high register, and not just because of the difference in pitch. Sampling an instrument in each of its distinct registers allows the synthetic version of the instrument to be more realistic.)

In addition to the aforementioned toilet flush sample, other samples used in “Water” include single drops of water, pouring water into a sink, and brushing my teeth. The extended water drop solos are a fugue of sorts, with the second voice entering with the same material but transposed up by a perfect fourth.

And there you have it – The Elements. Pretty “suite” music indeed! (Yes, unfortunately your reward for making it to the end of this lengthy post is a lame pun.)