composed entirely out of sine waves (plus a reverb)

-Blake

Can't all sounds be "composed entirely out of sine waves"? Isn't that what the Fourier transform is all about?

Good questions.

In a nutshell... Yes. As a general rule, all sounds can be composed of sine waves. The fourier transform breaks down a waveform into a series of sinusoids, each with its own frequency, phase and amplitude. The sum of these sinusoids equals the original waveform. I'm guessing you already know this.

Although I am using sine waves, the difference here is that there is no original waveform. Every bleep you hear is constructed entirely from scratch using sine waves. This is called additive synthesis in the world of sound design. Breaking down vs. Building up.

As for the reverb... After the bleeps are generated, they are sent through the reverb instrument to spatialize the sound. Hope I've answered your questions.

Ah, got it. So its the difference between "The Brooklyn Bridge consists of individual atoms" and "The Brooklyn Bridge was designed and built at the atomic level from individual atoms." The former is trivially true, the latter is quite a different story.

-Blake

Personally I'd be tempted to post the comositions into Perl Poetry, even though strictly speaking, the sounds aren't "composed of the Perl language". But they do sound like strings of pearls.

Matt

My intent was to make the rhythms to sound as natural as possible. I love listening to the patterns the rain makes, and I made my best attempt to emulate these patterns. I used a simple chaos equation for choosing rhythms because I think it sounds more natural than the rand function. Chaos is also very motivic.

If it sounds like jazz, it's probably because I went to jazz school for 4 years. :) Although I studied electronic music instead.

++jake Way cool! I like the surprise at the final best. This really sound crystal clear on my machine..

I wonder, did you compose this in a traditional manner ("That's what I want to hear, how will I get it?"), or are parts of this random/formulated/calculated ("Let's see how this sounds?")?

I'd love to hear some details on the approach you took in composing this (but maybe that does not belong here?

So long,
Flexx

A little of both. When I was designing the instrument, I definately took the "Let's see how this sounds?" approach. I did a lot of experimenting generating csound code with perl. After my experiments, I had a bunch of material that was ready to be made into the piece. Rain was part of this material.

When I was ready to compose Rain, the instrument had already been developed. I took my Rain instrument and made it a perl sound object (Rain.pm).

But as for composition goes, it was very much a "That's what I want to hear" type thing. I spent a long time fine tuning the sections of the piece and rewritting code to get it do pseudo-exactly what I wanted to do. I say pseudo-exactly because it doesn't generate the exact same csound score everytime. But it always generates within the limits of where I want everything to be.

I should also point out that the Rain sound is based on something I call a Harmonic Tree.

The whole thing took 9 months to complete. But to be fair, I wasn't working on it the entire time.

I am a musician myself (voc, perc), but I am not a synthesizer specialist. Howerver, AFAIK, in the early days of sound synthesis, all sounds were derived from simple waveforms (like sine, square, triange, chainsaw (like triangle but with a steep upward/downward flank), and so called envelopes (you could call them amplitude operators), which were "imposed" on those waveforms. This envelope was described by four parameters: attack, decay, sustain, and release (or short ASDR).

• A (time to reach maximum amplitude from keypress)
• D (time to drop to sustain amplitude after reaching max amplitude)
• S (amplitude to keep as long as the key is pressed)
• R (time to reach zero amplitude from S amplitude after key release)

AFAIK, A and S were linear, and the D and R parts of the envelope where parabolic.

  A       D       S      R
       o
      o o
     o   o
    o       o
   o           oooooooo
  o                   o
 o                     o
o                         o

Later 6 or more parameters where usual. Most of the modern soundcards still have ADSR/FM synthesizers build in. A famous specimen of these was the Yamaha OPL-3 chip all SoundBlaster cards had on them.

I hope this is somewhat interesting, ;)
so long,
Flexx

Links I Like: csounds.com, Csound Magazine, Buzz Tracker, Harmony Central, musicdsp.org

Books I Like: Computer Music (a must), Computer Music Tutorial, The Csound Book

It wasn't additive synthesis. The technique you are referring to (i think) is Linear Prediction Coding (LPC). It's the same thing that was used in technologies like the Speak'n'Spell and Dr. SBaitso. You may also remember the voice of Joshua in Wargames.

The entire process works in three stages:

Perl -> Csound -> Soundfile

Perl is the top level in this chain. It's where I compose the piece. Rain.pl generates Rain.sco. Csound is a synthesis language. Csound takes Rain.sco and Rain.orc and generates a soundfile.

As for the errors, I'm not getting those on my machine. But I take full responsibility for anything broken in my code. :)

Nothing broken here, the message is just a warning, not an error. The reason it does not occur on your machine might be that you're using differnt perl builds.

So long,
Flexx