Direkt zum Inhalt

In the World of Grains – Part 9

granular synthesis

In the World of Grains – Part 9

(contains embedded video – excerpt from my e-book about sonic grains: https://www.dev.rofilm-media.net/node/332). And if you want to support my work, please make use of the “PayPal” button at this website – thank you!)

A lot of parameters of single grains, and a lot of parameters of (complex) clouds!

But let´s be practical: a lot of these parameters, of these properties will be redundant, with very short grains, because the differences in – e.g. - the development of the spectral content of a grain during its lifetime won´t be noticeable, if the grain´s length is getting too short.

granular synthesis

This aspect is worth some investigations. Single grains first. Meaning: I explore the sonic behaviour of exactly 1 single segregated grain.

In 11 test series, and all in all 275 tests I found the following overall results:

The higher the frequencies of the grain´s sonic content are, the shorter can the duration of the grain be without the listener losing the ability to determine its exact pitch.

And it´s the same with the listener´s ability to recognise changes in the grain´s pitch, meaning recognising, THAT the pitch changes AT ALL, even if not being able any more to tell HOW the pitch changes. The higher the frequency content is, the shorter can the grain be, and we still notice CHANGES of pitch.

None of these two first findings should astonish us, because higher frequency means more cycles per lifetime of the grain.

If we the grain in question is not segregated, is not repeated with gaps between the repetitions, but with consecutive repetitions overlapping each other, the grain can even be shorter.

The shortest lifetime, that still enables us to recognise, if or not the frequency is changing, or even to determine the pitch, is not only dependent on the frequency region of the grain´s content (higher of lower), but also depends on the KIND of sonic content (waveform etc.).


A grain, which contains only a simple sinewave of 1,000 Hz must be at least 17 milliseconds long, to enable me to recognise its exact pitch, whereas a grain containing the sound of a clarinet, can be remarkably shorter. It needs only a duration of 8 ms, and I still can recognise its pitch. With some complex synthesized sounds the grain´s length could go even further down. With pulse waves of a duty cycle of 20% I needed the grain to last at least 11 seconds to determine its pitch.

I´ve tested frequency bands from 62 Hz up to 8 kHz, with the following kinds of sonic content: sine, triangle, pulse (duty cycle = 20%), square (duty cycle = 50%), saw, synthesizes wave with complex spectrum, white noise, a clarinet, human voices (children), wildlife (birds), machine sound (motor of a washing machine). In my book (see above) You´ll find the detailed results of all these tests in Appendix A. Please take them as a general guidance. The absolute numbers will vary a bit, as the sense of hearing is highly individual. Please follow the link to the video showing a bit of how I did these tests:


So far constant frequencies. Now for changes during the grain´s lifetime – we could call them “inner” changes. I´ve tested two frequency jumps: 12 semitones and 6 semitones. And again I tested different kinds of sonic content.

How long does the duration, the length, the lifetime of a grain have to be to enable me to recognise frequency changes of its sonic content?

First of all: we need remarkably longer grains to recognise frequency changes in the grain, than we need to recognise frequency changes between different (e.g. consecutive) grains. And the grain´s length must be even longer to recognise these changing frequencies of the grain´s content.

We could have expected, that bigger differences between these frequencies in the grain make it easier to discover, that the grain´s content is changing. But it seems, that with some kinds of sonic content it is not the frequency, which is important whether or not we recognise changes, but the overall sound, the spectrum.

With the sound of a clarinet around 1 kHz it was a little bit easier to discover frequency changes of only 6 semitones, than those of a whole octave. This phenomenon was even clearer with the complex synthesized sound at around 500 Hz.

Sure, we know that acoustical instruments as well as certain synthesized sounds change their sonic character more in certain frequency regions, and less in others (because of the physical behaviour of the instrument, or – with synthesised sounds – because of higher partials vanishing into the region of inaudibility). But this phenomenon deserves some more detailed and broader investigations in later editions of this book.

Interesting enough it´s the triangle wave, and the pulse wave, which only need the shortest grains to enable us to recognise sonic changes of their content. In my book (see above) You´ll find the detailed results in Appendix B. Now for clusters of grains. Well, that´s quite simple now. We can take the figures from the tests with single grains, but have to reduce the mentioned minimal grain durations by a factor, which depends on the density of the cluster, and/or on the frequency of repeating a single grain (which is – strictly taken – not a cluster).


The more grains of the same sonic content are overlappingly repeated, the more the minimal length (duration, lifetime) of the grains can be reduced, with the parameters bandwidth (and slope) and center frequency (and slope) being still of sonic meaning.

The video shows how I did the testing:


I´m going to dive deeper into the details of this matter (clusters) in the next part of this series of ebooks about granular sound processing.

to be continued

to part 1: ("A Short History of Granular Synthesis - Part 1"):https://www.dev.rofilm-media.net/node/340

to part 2: ("A Short History of Granular Synthesis - Part 2"): https://www.dev.rofilm-media.net/node/342

to part 3: ("A Short History of Granular Synthesis - Part 3"): https://www.dev.rofilm-media.net/node/346

to part 4: ("A Short History of Granular Synthesis - Part 4"): https://www.dev.rofilm-media.net/node/356

to part 5 ("In the World of Grains - Part 1"): https://www.dev.rofilm-media.net/node/364

to part 6 ("In the World of Grains - Part 2"): https://www.dev.rofilm-media.net/node/373

to part 7 (“In the World of Grains – Part 3”): https://www.dev.rofilm-media.net/node/378

to part 8: (“In the World of Grains – Part 4”): https://www.dev.rofilm-media.net/node/385

to part 9: (“In the World of Grains – Part 5”): https://www.dev.rofilm-media.net/node/390

to part 10: (“In the World of Grains – Part 6”): https://www.dev.rofilm-media.net/node/398

to part 11: https://www.dev.rofilm-media.net/node/407

to part 12: https://www.dev.rofilm-media.net/node/414

to part 14: https://www.dev.rofilm-media.net/node/486


Add new comment

Das Sprachkürzel des Kommentars.


  • Keine HTML-Tags erlaubt.
  • Zeilenumbrüche und Absätze werden automatisch erzeugt.
  • Website- und E-Mail-Adressen werden automatisch in Links umgewandelt.