In addition to the mp3's on this page, there's Images to Sound and Integer Sequence Noise.
For the last several years, my main sound output has occurred during February, for the RPM Challenge.
Not a very successful RPM Challenge 2013. I really wasn't very motivated, and I only got three tracks acceptably made. But I like these three more than a lot of stuff I've made.
1. Simple Composition 1 - 2:13 - A simple composition idea. Basically, three instruments play a loop of three notes, with rhythms in a four note cycle, so the combinations of notes at any one instant vary throughout. There are some changes in rhythm, too, by shortening the scale of the rhythms, and some timbre changes. Actually harmonious.
2. Simple Composition 2 - 4:02 - Another one based on the same principal, but with more dissonance, and a little filtering to give the timbre a little more variety.
3. Triangle Composition - 4:00 - Ten balls move linearly at various constant speeds, bouncing off the sides. Every triangle formed by these balls is drawn, the color determined by how equilateral the triangle is (redder = more equilateral). A sound event is generated for each triangle, 60 times per second, determined by the size, location and equilaterality (?) of the triangle. There is an accompanying video.
All tracks recorded during February 2012 for the RPM Challenge 2012.
You can download all tracks and the liner notes as a zip file.
You can download the liner notes as a pdf which you can fold and glue into a CD envelope.
I was really busy this February, with drawing, cycling, and too many cheating students in my classes, so my ideas were pretty scattered. I think this was successful, though, in the fact that I got a few new techniques tried, and an old one or two expanded on.
1. Temporary Rhythms 2 - 5:10 - This is a followup to last year's Temporary Rhythms. The idea is the same: a set of random rhythms (on a 16 beat measure) are generated, and gradually combined in different combinations and with different "instruments". For this one, I also layered half-speed and quarter-speed versions of the rhythms, too, so it's nice and thick, and it gets thicker as the track progresses, but I wish I'd pushed it a bit more. Next year I think I'll try something similar, but with more extreme sounds.
2. Granular 1 - 8:20 - This is a pile of one-wavelength sine wave sound "grains", thrown out on the timeline with varying densities and varying pitch centers. I was hoping for a little more variety in the sound, but I think the rain-like nature and the roaring bass parts are good.
3. Sines Inside Sines 1 - 1:00 - A very old idea which I have recently
come back to. I recently found how to write sound files with a Perl script, and this
allows me total control of the final result, plus it gives me access to proper
trigonometric functions (which I have had problems with in Csound). So, the idea is
simple: create a sound file from a straight-up mathematical function.
For this track, I define three functions of the form
and then the output waveform is essentially f(t)=f1(t) f2(t) f3(t).
This track turned out nicely crunchy. The challenge with this kind of stuff is to get the sounds to be interesting without going beyond the limits of the 44100 samples-per-second system (essentially, you have to watch that the derivative of the functions do not get too great).
4. Frequency Modulation 1 - 5:02 - This is perhaps my favorite track on the CD.
This again was created in Perl, with functions of the form
which goes twenty deep. The function you hear in this track changes periodically, more and more rapidly until the last 90 seconds or so when it changes at a constant rate. I really like the types of sounds this method generates, though I haven't played with it long enough to find anything that changes nicely over time by itself.
5. Modular Arithmetic, Mostly 1 - 3:20 - This was also created in Perl. The idea was to use no standard oscillator-like components, like sinusoidal functions. This was created entirely with integer operations. Essentially the i-th sample in the track is determined by testing i as an integer against various moduli and conditions which themselves change as the track progresses. I'm really happy with the unexpected nature of this track and some of the timbres are just great, and particularly the change around 31 seconds in.
6. All Together - 8:00 - Created with Csound (QuteCsound actually). I had been really enjoying the sounds I was getting in QuteCound just by letting five or six oscillators interfere with each other, so I set up a bunch of oscillators with linearly changing frequencies interfere with each other, mixing and combining them in various ways and let it go. This was the last track I made this year, and I probably made it a little longer than it needs to be but I like a lot of the timbres, and the blatant "electronic" feel of it.
7. Sines Inside Sines 2 - 5:00 - Another like track 3, but with a much more mellow feel. Same function type, with different coefficients and more of them.
2011 was the first RPM Challenge I completed! I made just over 36 minutes of sound in nine tracks, and they don't all completely sound awful. Here they are:
This is a basic rhythm composition using an idea I've been working for a long time. Take snippets of speech, say, of fixed length (in this piece, 0.9 seconds). These are taken at random from a longer sample. Each snippet will be looped over and over, so there is the rhythm resulting from the fixed length of the snippet, but then there is also rhythm caused by the content of the snippet. By overlaying these loops, and starting and stopping them at various times, different combinations of rhythms result.
In the middle of the track, ring modulation is applied to some of the loops to give a more colorful sound.
The sample is taken from a record I have which apparently was given to folks considering buying a boat as part of the boat company's marketing. It seems that owning a boat involves some kind of pride, along with enjoyment.
I struggled with this compositional idea all month. I think there is more to get out of it, but this is as good as I could get in February.
The idea is to have the spectrum of the sound at any instant be a function of the spectrum of the sound at the last instant. The spectrum is a set of amplitudes at 50 fixed frequencies (I will definitely use more in the future). Every 0.01 seconds, these amplitudes are modified according to a function which depends solely on the set of amplitudes. This is similar to a one-dimensional cellular automata (the difference being that in a CA, there tend to be a finite number of "states", but here the states are the amplitudes, so they are, theoretically, infinitely variable).
I think I need to go back to some basic CA methods, and see what kinds of sound I can get out of them, and then gradually add complexity, so that I can get a better idea of what is possible.
This is similar to the last track, in that the spectrum at each instant is a function of the spectrum at the previous instant. Each spectrum is treated as a vector, and this vector is multiplied by a fixed, randomly generated matrix to create a target vector and a weighted average of the original vector and the target vector are created. This weighted average vector is the next spectrum.
If this process is allowed to continue, the result will eventually be noise, as the spectrum moves asymptotically toward having all equal amplitudes. So, occasionally, parts of the spectrum are zeroed, or otherwise changed, to create a more varying sound.
The Csound score file for this is the largest I've ever produced. It is 1400 megabytes.
This started as a dare. Some time in February, Jenni and I were walking to work. A man driving a truck stopped to let us pass, and waved a finger in our direction as some kind of communication. I suggested that he was envisioning a tiny rodeo: I mimicked him saying "Ya-Hoo!" as he waggled his finger in a circular fashion, and made up the song on the spot: three descending chords plus the lyric "Finger Rodeo" over and over. I said it should be my next RPM track, and Jenni dared me to do it, and so I did it.
Two vocal layers, one guitar, and then massive manipulation via vocoders I wrote in Csound. The classic vocoder uses synthetic waveforms to approximately reconstruct an audio signal. With a small number of waveforms, a robot-type voice results. This is the effect I applied to the guitar. For the vocals, I used filtered noise instead of nice waveforms.
Various looped rhythms which start and then end autonomously. I worked on this to get the various waveforms to be nicely varied, and, with the addition of reverb here and there, I think it has an okay sound.
My welding teacher told me about an idea she had to modify music boxes by grinding off the little nubs on the cylinder, and then adding nubs with solder. This got me thinking of other ways to modify them. I got four music boxes (which all play theme music from popular soap operas) and removed the steel comb and turned it upside down, so that high notes would be low and vice versa. I learned when I tried this that the comb is shaped so that playing it upside down is not as easy as in the normal direction: the comb is asymmetrical. Consequently, the sound is extra plinky. To hold the comb down, I had to use binder clips, since the screw holes are also asymmetrical. The binder clips add additional resonating.
Each box was recorded separately, and this track is all of them mixed together. The music boxes have a flywheel to smooth out the power transmission from the wound spring to the cylinder, and you can hear the humming of the flywheels on this recording.
The same idea as the first track. Here the snippets are longer, and there is no ring modulation in the middle. This recording of FDR has more interesting sounds in it than the sample for the first track.
This is one of my pet ideas: the spectral inversion of sound. For this track, I took an old recording I made playing acoustic guitar, and modified it using Csound's phase vocoder capabilities. I then messed with the phase vocoder during resynthesis to essentially invert the spectrum over the course of the track: high frequencies become low, and vice versa.
An old sample from Dragnet about the effects of computers on labor.
Another idea I have been trying for a long time is comparison. Considering two samples (or other waveforms) as functions of time, f(t) and g(t) say, we can create a new waveform such as h(t)=max(f(t),g(t)). This often produces quite a mess. Here, the Dragnet sample is "maxed" with a few fairly low frequency oscillators. As the oscillators accumulate, the voice is drowned out, which I thought amusingly metaphorical (or metaphorically amusing).
Four Things (5:23, stereo, mp3)
A type of dessimation of a sample, spread over 5 minutes. For RPM'10.
Avocado Pesthole (4:00, stereo, mp3)
Thicker Csound sounds than my usual. For RPM'10.
Yesterday Starts with Yes (5:03, stereo, mp3)
Another dense Csound thing. For RPM'10.
Bliptronic suite (7:43, stereo, mp3)
My brother Chris gave me a Bliptronic 5000 for Christmas, and I used it to create this triptych of little pieces.
Two tracks from the '10 RPM Challenge. Made with Csound. The spectra are made from Pascal's triangle modulo 2 or 5 (i.e., cells of Pascal's triangle, viewed horizontally, which are congruent to 0 or 2 mod 5 correspond to sound events).
cellular automata spectrum (60 seconds, stereo, 0.9 MB mp3)
This was one of two tracks I made for the '09 RPM Challenge. Made with Csound. The idea is the the sonogram of the track is a one-dimensional cellular automata. Each iteration of the CA corresponds to a short span of time, which gets shorter as the track progresses, just to give the thing some kind of change. I think I had 100 different frequencies, each of which correspond to a cell of the CA.
infamy oscillators (60 seconds, stereo, 0.9 MB mp3)
This was one of two tracks I made for the '09 RPM Challenge. It's really noisy. Made with Csound.
los rhythm 01 (240 seconds, stereo, 5.5 MB mp3)
This track is generated with Csound and Perl scripts. I finally got around to writing a perl script that could look in a directory for samples, and create a csound orchestra file with one instrument for each sample. This makes it much less tedious to add samples to a piece.
In this track, I was going for a kind of Steve Reich/Esplendor Geometrico gradually changing tempre/rhythm kind of thing. I think it came out okay, but since it's all the product of code, it is forever tweakable, which is something I really like. (added 3/1/08, created during RPM 08).
100 tones (300 seconds, stereo, 6.9 MB mp3)
This is 100 sinusoidal oscillators with slowly, linearly changing frequencies, all spread across the stereo image. Kinda fun, I think. (added 3/1/08, created during RPM 08).
Simple Csound Quadrafuzz (26 seconds, stereo, 0.5 MB mp3)
For years I've wanted to try the "quadrafuzz" idea out in Csound. The Quadrafuzz is a device by Craig Anderton that splits the input signal into four bands, distorts each one, and mixed the four bands back together. This is very easily done in Csound, though here I've used pure clipping for the distortion, which some may find a bit "harsh". The example mp3 is the result of processing a sample of a "clean" electric guitar with this Csound instrument. Here are the orchestra and the score files (added 12/28/07).
Simple Csound vocoder example (48 seconds, stereo 1.1 MB mp3)
I've wanted for some time to make a simple "robot voice" vocoder effect in Csound. This mp3 contains an original sample, followed by a 40-band sine wave vocoded version, then a 40-band sawtooth wave version, then a 10-band sine wave verions, an upside-down version and then a 40-band filtered noise band version. The key to all of this is Csound's balance opcode. Here are the Csound orchestra and score files.
The output bands are split to the left and right channels, so though the original sample is mono, the output is quite stereo.
Beethoven Ends (63 seconds, stereo 1.5 MB MP3)
I always enjoy the last minute or so of Beethoven symphonies, so I made this piece: it's the last minute of Beethoven's 1st, 3rd, 4th, 5th, 7th, and 8th symphonies, all played at the same time, ending simultaneously. I recommend headphones, or a subwoofer.(Uploaded 9/4/2007)
Rhythmic Noise 1 (147 seconds, stereo 2.2 MB MP3)
A "rhythm track" made with only filtered noise. There is no real intentional rhythm besides a basic quantization of time. The rhythms you might hear are all accidental, caused by the varying of the filtering according to sinusoidal functions.
Short Noisy Granular 1 (62 sec. stereo, 972k MP3)
A short, noisy bit of granular composition.
The score is huge: the sound file is made up of thousands of
tiny sound events.
I think the method has some potential.
Divisors, to 210 (81 sec. stereo, 952k MP3)
I was wondering what this would sound like for some time, and finally got around to making it. This is a sound representation of the small prime divisors (2, 3, 5, and 7) of the numbers from 1 to 210 (210 being the product and theorefore the least common multiple of these primes). It is essentially just counting, but instead of hearing the usual name for each number, you hear the names of each number's small prime divisors. Meh. (Recorded 8/18/03)
The Sound of Primes (60 sec. mono, 468k MP3)
This file was generated
to give an audio representation of the distribution of the prime
numbers. If p < 6000 is a prime number, then there is
a bit of sound between time p*0.01 and (p+1)*0.01 seconds.
Each bit of sound is a sine wave with frequency 100*k hz, where
k is the gap between p and the previous prime, with
a trapezoid thrown on to make sure there is no clicking.
Thus all the primes less than 6000 are represented.
I think the best thing about this piece is that it gives a real feeling of irregularity. We hear the unpredictability of the primes.
Brownian Tones (3 min. stereo, 3 meg MP3)
This piece was created by layering several random walks of tones. The tones are simple sine waves damped by trapezoidal envelopes. They each move randomly from their own starting frequencies to their final frequencies. These are not symmetric walks: some walks increase in frequency, some decrease. As the walk progresses, the tones overlap, and since the walk is slow and the frequencies similar, dissonance results.
Her Voluptuous Body (2.5 min. stereo, 2.3 meg MP3)
Listening to Steve Reich's 'Early Works' CD inspired me to make this piece,
and Csound made it very easy to do. I started with a sample of Maria de
Medeiros saying "her voluptuous body". This is looped and played over and
over for the entire piece. Let us say that it is played n times.
Then, another loop is made by taking that original sample and adding blank
space to the end of it so that, when looped and played for the entire
piece, it is heard n-1 times. Similarly, there are loops for n-2, n-3, n-4
n-5 and n-6. Each loop has its own stereo position.
The end result of all of this is that at the start, all of the samples are "in phase" with each other. As the piece progresses, they fall out of phase, and then, in varying combinations fall back into phase with each other, creating interesting timbres and rhythms. At the end, the samples all come back into phase again. I wrote a Perl script to generate the Csound score for this, so it's very easy to create thicker, or thinner pieces with any samples, and of any length.
Natural Thing (3.5 min. mono, 1.6 meg MP3)
An old audio piece (original made 8-bit, 22khz, on a Mac Quadra), edited together from a 50's sex education record and a plumbing installation flexy-disk. I've discovered in the process of this that record crackle sounds really bad in mp3 format - oh well. Recorded 1995 or 1996.
Back in the late 80's, I recorded some stuff under the name "Bop Apocalypse".
Going Around the Block (3:53 stereo(?), 5.3 meg MP3)
Recorded in 1989 on 4-track cassette, and released on the RRRecords compilation "Art is the Handmaid of Human Good" (on vinyl!). This was a compilation of songs about Lowell, Massachusetts, where RRRecords is located, and where I was a student at the University of Lowell. So, I had to make a song somehow about Lowell. This is what I came up with.
Exploited (6:13 stereo(?), 8.5 meg MP3)
Recorded in 1990 for a four cassette RRRecords/Statutory Tape compilation, "The Lowell Plan (Planned Obsolescence)" (Statap 015), of songs by Lowell artists (though not necessarily about Lowell). I can't remember the motivation for this one.
Clapping (2:39 stereo, 3.6 meg MP3)
Recorded in 1989 or 1990 on 4-track cassette. Samples from Vanna White's autobiography, and lots of rhythm and noise.
The Most Popular (6:18 stereo, 8.7 meg MP3)
Recorded probably in 1989 on 4-track cassette. Nice samples from a Disney record about Isaac Newton that I still have. My favorite part: a little girl says "I've never been in a cave before" with two different inflections.
Masterpiece Bologna Sandwich, by The Cretins (6:13 stereo, 6 meg MP3)
Recorded in 1987 or 86 with Tim Shea. The WJUL (now WUML) is located under library on campus, and there is a concrete tunnel you go down to get to it. This was recorded in the tunnel to get all the nice reverb, recorded in stereo. Tim's playing bass and does all the vocals on this, I mostly just banged on things. Recorded on 1/4-inch reel-to-reel. We called ourselves The Cretins.
Terminally Ill Fratboy, by Puke (1:55 stero, 2.8 meg MP3)
Recorded at WJUL with several friends in probably 1988 or 1989. This is a parody of the song Golf by Art Phag. It includes some sound effects from records the station had; I believe the primary one was labeled "Viking raiding party" or some such. I played guitar on this, carefully put out of tune. Recorded on 1/4-inch reel-to-reel.
Dental Damnation, by ? (2:02 stereo, 2.9 meg MP3)
Recorded at WJUL with two friends in probably 1989 or 1990. The percussion was an Alesis HR-16. The rhythm is an amusing thing here, getting progressively more irritating. Recorded on 1/4-inch reel-to-reel, and I must say, it sounds quite good.