Granular Synthesis 101
Each of Camel Audio Alchemy’s four oscillators can perform granular
synthesis. The relevant parameters for granular control are the Position
and Stretch knobs in the top row; the Size, Dens (density), RTime
(random time amount), and RPan (random panning amount) knobs in the
bottom row; and the Window drop-down. |
Musicians who use synthesizers have an amazing smorgasbord of sound design tools to work with: analog, FM, sample playback, physical modeling, wave sequencing, and other technologies are all found in a variety of instruments. Granular synthesis is not the most popular technique, but it’s a terrific resource for producing floating clouds of sound, rich animated pads and drones, pitch shifting, madly stuttering vocals, and glitch textures.
Several software instruments, including Camel Audio Alchemy (see Figure 1) and Native Instruments Reaktor (see Figure 2), have excellent granular synthesis implementations. Using experimental freeware such as Csound and Pd, you can take the granular experience even further. A few granular-based plug-in effects processors are also available.
What’s in a Grain? In granular synthesis, the sound you hear is built up out of dozens, hundreds, or thousands of short sonic elements called grains. In most granular sounds, the listener doesn’t perceive the grains separately; they tumble on top of each other to make a smooth texture, not unlike how a beach is made up of grains of sand. (Using larger grains and separating them from one another is also a useful technique.)
Grainstates SP, one of the instruments bundled in Native Instruments
Reaktor, can load a sample file and sequence up to eight different
granular “snapshots,” each of which provides control over the pitch
jitter, pitch shift, length, and so on. The three knobs along the bottom
of each snapshot control the crossfades between settings as Grainstates
moves from one snapshot to the next.|
In the early days, primitive forms of granular synthesis were accomplished using analog tape decks, but these days, all granular synthesis is digital. The source material for granular can be any digitized sound, from a sine wave to a recording of people speaking. Sampled sounds generally produce richer and more evocative results than simple waveforms.
If you’re using a spoken word or instrument phrase sample as a source, you should truncate the start and end so that the whole sample contains sound. Compressing and normalizing the source file would also be a good idea, so that the dynamic level is fairly uniform.
It would be impractical to try to specify the characteristics of each grain individually, so a granular synth will give you higher-level macro parameters. Using these parameters, you can dial in the details of all of the grains at once. By modulating one or more of the parameters while the sound is being generated, you can easily conjure up a granular texture that changes over time.Dimensions of Control
The parameters you’ll be able to control in a granular synthesis engine include the source sound material to be used; the type of amplitude envelope the grains will have; the lengths, amplitudes, pitches, and pan positions of the grains; the playback start point for the individual grains within the source sound file; and the density of the granular texture—that is, how frequently new grains start sounding.
In some instruments, the amplitude, pitch, panning, and start point can be varied randomly if desired. The amount of randomness is also an important parameter. For instance, by starting with a tone that has no pitch randomness and then increasing the pitch randomness, you can create a pitched pad tone that evaporates into a cloud of noise.
With a sampled source sound, you will probably be able to control the start point of the grain within the sample. If the sample varies, moving the start point (forward or backward) from one grain to the next will cause the granular tone to evolve. This technique is extremely useful. If the start point moves through the sample more slowly than the sample was recorded, the sample will be time-stretched. When the start point is stationary, a sample becomes essentially a bank of static waveforms that can be used in subtractive synthesis.
The envelopes of the grains are often specified using a parameter called the window type (see Figure 3). Windows with squared-up rise and fall times will add noisy high frequencies to the tone; windows with gradual rise and fall times will give you a smoother tone. In some implementations, such as when using Csound’s granule opcode, you may be able to control the rise and fall times separately.
|Fig. 3. The Hann window is one of several mathematically derived window types used as amplitude envelopes for individual grains.|
The density and grain length parameters work together to determine whether individual grains can be heard in a sparse texture, or whether the grains overlap, blending together into a smooth tone. When the grains overlap, the overall sound will be louder, and you may also hear phase cancellations as the waveforms of the overlapping grains blend together.
By moving the start points of successive grains through the sampled source material at something like real time (the speed at which the source was recorded) while shifting the pitches of individual grains up or down, granular synthesis can perform a type of pitch-shifting.Ear Food
To hear a few examples of granular synthesis tones, visit emusician.com/november2013
. The ideas in these sound clips may be more useful for experimental music than for mainstream pop styles, but pop producers are always looking for fresh sounds, and granular certainly does fresh.
Jim Aikin has written hundreds of product reviews and tutorials for Electronic Musician and other magazines over the course of more than 30 years. His books on music technology include Power Tools for Synthesizer Programming (Hal Leonard Publishing) and Csound Power! (Cengage Learning).