What is granular synthesis?

Wikipedia has this to say on the subject of Granular Synthesis:

Granular synthesis is a basic sound synthesis method that operates on the microsound time scale. It is often based on the same principles as sampling but often includes analog technology. The samples are not used directly however, they are split in small pieces of around 1 to 50 ms (milliseconds) in length, or the synthesized sounds are very short. These small pieces are called grains. Multiple grains may be layered on top of each other all playing at different speed, phase and volume.

The result is no single tone, but a soundscape, often a cloud, that is subject to manipulation in a way unlike any natural sound and also unlike the sounds produced by most other synthesis techniques. By varying the waveform, envelope, duration, spatial position, and density of the grains many different sounds can be produced.

The result is usable as music, sound effects or as raw material for further processing by other synthesis or DSP effects. The range of effects that can be produced include amplitude modulation, time stretching, stereo or multichannel scattering, random reordering, disintegration and morphing.

Okay, what is a grain cloud, then?

Again, from Wikipedia:

In music a cloud is a sound mass consisting of statistical clouds of microsounds and characterized first by the set of elements used in the texture, secondly density, including rhythmic and pitch density (Roads 2001, p.15). Clouds may include ambiguity of rhythmic foreground and background or rhythmic hierarchy.

Why would I want a grain cloud?

Grain clouds produce wonerfully interesting textures from the most mundane of source material. Take any recording you have laying around, plug it into Atomic Cloud, and you'll be amazed at what you get back. If you're interested in synthesis or sound design, it's hard to ignore the sonic possibilities of grain clouds.

What does Atomic Cloud sound like? Can I hear a little bit before downloading?

Sure, here are some demos:

Guitar: original source file | grain cloud 1 | grain cloud 2

Didjeridoo: original source file | grain cloud

Chimes: original source file | grain cloud

Is Atomic Cloud a standalone application, or a plugin (VST, RTAS, AU, DX)?

Atomic Cloud is a standalone application, and does not require anything additional in order to run. You use it just like you would use any other application, usually by double-clicking on the application icon. Once you've generated a grain cloud that you like, you can render it to disk using Atomic Cloud, and then bring your grain clouds from Atomic Cloud into a host application, such as Cubase, Logic, or Pro Tools, but Atomic Cloud does not presantly run as a plugin of any sort.

Why doesn't Atomic Cloud run as a VST plugin?

Getting a really smooth cloud can take quite a lot of CPU power. Basically, the more overlapping grains you have at any given time, the more CPU cycles Atomic Cloud will use. Losing the overhead of the VST host and any other VST plugins means you can get a much denser grain cloud on the same system, so I decided to make Atomic Cloud a standalone program.

What operating systems does Atomic Cloud run on?

At the moment, Atomic Cloud is Windows-only.

How do I use Atomic Cloud?

Atomic Cloud is easy to use. After starting the application, click the "Browse" button, and select an audio file to use as the source material for the grain cloud. Atomic Cloud will load most uncompressed audio file formats. The demo version of Atomic Cloud is limited to using 10 second or shorter source files, but the full version can load audio files of any length to use as source material. Playback will start automatically, and can be stopped and started with--surprisingly enough--the stop and start buttons. Before looking in depth at the parameters, it is important to understand that Atomic Cloud employs a playback head, just like a regular audio editor. The Scan Rate controlls the speed of this playback head. At a neutral Scan Rate of 50, the playback head will scan through the audio file at real-time. Grains are generated from the source material found at the playback head. However, the Jitter parameter introduces randomness into the equation. As you increase the Jitter from 0, grains are generated at random distances farther and farther away from the playback head. What this basically means is that at a Scan Rate of 50, and a Jitter of 0, with a grain rate and grain size set high enough that you don't hear any space in between grains, you will essentially hear the file as if it were being played back in a normal audio editor. This effect can be heard by choosing the Normal Playback preset, and I suppose, could be considered the most neutral of settings. Start adjusting the controls away from this in order to get more interesting effects.

Parameters on the GUI are as follows:

  • Start and End Location: You can load an audio file of any length, but you might only want the grain cloud to be generated by a small portion of the file. Use the Start and End Location to constrain the playback head by setting limits on what portions of the audio file can be used to generate the grain cloud.
  • Grain Rate and Grain Size: The Grain Rate controls how quickly grains are generated. Low settings produce very few grains per second, whereas high settings produce many grains per second. The Grain Size controls how long each grain is. High grain rates and long grain sizes are capable of generating remarkably smooth textures, but also tend to take quite a bit of CPU power. You will likely not be able to turn both the Grain Rate and Grain Size up all the way without running out of CPU cycles.
  • Buffer Rate: The rate at which each grain is played back. This essentially has the effect of altering the pitch of each grain.
  • Scan Rate: How quickly the playback head moves through the audio file. With low levels of Jitter, grains are generated near the playback head.
  • Jitter: How closely the generated grains are constrained to the playback head. A very low jitter setting, with a very high grain rate has the effect of playing the file back pretty much as it would sound regularly. High jitter settings randomize the sound greatly.
  • Amp Level: Since any number of grains can be generated at a time--depending mostly on your grain rate and grain size choices--it's quite easy to get digital clipping. The built-in limiter controls this, but generally speaking, higher grain rates and grain sizes should often be accompanied by lower amp levels, so as to avoid clipping or excessive limiting.
  • Clip Indicator: Unsurprisingly, the clip indicator indicates when digital clipping has been detected. Click it to clear the indicator. Clipping should really only occurr if you've turned off the built-in Limiter. In this case, you should turn down the Amp Level in order to avoid clipping, or turn the Limiter back on.
  • Limiter: Atomic Cloud is equipped with a built-in Limiter. This comes in handy, because long grains are likely to overlap, so it is very easy to generate a grain cloud that starts clipping if you don't watch the gain. The best solution for this is to turn the gain down until you don't have any clipping, but the built-in Limiter will watch the gain for you, and turn things down when necessary.
  • 16/24 bit Playback/Recording: Playback and recording can occur at either 16 or 24 bit, regardless of the bit rate of the source file. 24-bit output is not available in the demo version of Atomic Cloud.
  • Automatic Sample Rate Sensing: The output sample rate is set to whatever the source file's sample rate is.
  • Buffer Size: The Buffer Size parameter mainly controls how smooth the amplitude envelope of each grain is and how much latency there is between when you move a control and when it affects the audio. As with all audio applications, a smaller buffer takes more CPU, but generally results in a better sound. In this instance, shorter grain sizes should generally be accompanied by smaller buffer sizes, since the effect of a large buffer will be most noticeable on short grains. However, larger grain sizes can be accompanied by larger buffer sizes in order to reduce CPU load without affecting the sound too terribly much.
  • Recording: Atomic Cloud can render the grain cloud to disk as you listen to it (note: recording is disabled in the demo version). This is the best way to record audio from Atomic Cloud, as the audio data is written straight to disk while you listen without any sort of DAC or ADC conversion. Audio files can be recorded in either 16 bit or 24 bit formats, and the rendered sample rate (i.e. 44.1kHz, 48kHz, 96kHz) will be the same as the source file. So, if you load in a 96kHz file, you'll get a 96kHz grain cloud.
  • Presets: Presets of your favorite parameters can be saved and called back at the bottom of the application. Saving and deleting of presets is disabled in the demo version.

Still need help?

Ask a question on the Sound Borb / Atomic Cloud Forum.