Music

Granular Synthesis

Curtis Roads is the master of granular synthesis (aka composing with small particles of sound) and I’ve been fortunate enough to study with him for the past few months. The resulting piece won the informal “People’s Choice Award.” It’s a study on sound transformation and gestural composition.

All of the sounds are manipulations of samples taken from the previous post, Player Piano Study. I put these samples into a granular synthesis instrument that’s controlled through a motion capture controller I’m working for my thesis. Depending on what parameters I specify, it creates really strange, mechanical sounds such as the ones you hear toward the beginning of the piece, or more mellow chirps such as the ones about two-thirds of the way through. It’s meant to be played in 4-channels (speakers).

[audio:http://amusesmile.com/old/sound/roadsMaster.mp3|titles=Pantograph]
Music

Player Piano Study

This piece is the result of experimenting with a Kawaii Digital Player Piano over the past few weeks. I used Renoise to create seven melodic sequences ahead of time and then performed the piece by activating different combinations of these sequences live. While recording I also improvised with the piano’s sustain pedal in order to build and fade intensity.

[audio:http://amusesmile.com/old/sound/Geiringer.mp3|titles=Geiringer Hall]
Code Music

Old Song, Old Language

Here’s a piece I dug up from last year when I was working quite a bit with RTcmix (Brad Garton’s awesome albeit antiquated musical scripting language). I like the piece’s meditative quality and how varied it sounds for being written with only a few lines of code. It doesn’t follow canonic form at all, but when I hear it I can’t help getting this impression.

[audio:http://amusesmile.com/old/sound/Canon_Mastered.mp3|titles=Canon]
maxamp = 1000
amp = maketable("line", 1000, 0, 0, 50, 1, 51, 1, 100, 0)

wavetable = maketable("wave", 1000, 1, 0.3, 0.2)
pan = 0.5

start = 0.0
start2 = 0.0
freq = 2000.0
freq2 = 0.0

for (i = 0; i < 100; i += 1) {
freq2 = freq/round(irand(1,6))
WAVETABLE(start, 4, 10000*amp, freq2, (.2 + random(0, 0.6)))
freq2 = freq/round(irand(1,12))
WAVETABLE(start, 4, 10000*amp, freq2, (.2 + random(0, 0.6)))

freq2 = freq/round(irand(1,12))
WAVETABLE(start2,  6, 10000*amp, freq2, (.2 + random(0, 0.6)))
start += 2
start2 += 3
MAXMESSAGE(0 \, freq2)
}
Art

Talking Heads: Speech Visualizations of the Past and Present

Last year I became interested in the possibility of comparing different politicians and public figures through computer assisted analysis of their speeches. Starting with archival videos of famous addresses, I created a small program that scans the volume level of the audio track and takes snapshots of the speakers at his/her loudest (i.e. most emphatic) moments. It then combines these snapshots into a single composite photo. The resultant images reveal otherwise hidden facial features and patterns of body language. When viewed in series they allow us to compare speakers within a much more controlled set of parameters than if we were simply to watch these videos side by side.

Last April I was fortunate enough to be invited to give a talk on this project at the Critical Themes in Media Studies conference in New York City. Unfortunately I never got around to creating a real post about the piece, however renewed interest in the topic has driven me to fix this. Below you will find image and sound examples taken from the talk. If you’re interested in using this program for your own work, the code can be found here. Over the next few weeks my close friend YuanYi Fan will be using the software to analyze candidates in the upcoming Taiwanese Presidential Election, so links will surely follow.

More examples and the code itself can be found on the Full Project Page

Benito Mussolini

[audio:/old/sound/musso.mp3|titles=Benito]

Adolf Hitler

[audio:/old/sound/hitler.mp3|titles=adolf]

Dr. Martin Luther King jr.

[audio:/old/sound/luther.mp3|titles=luther]
[audio:/old/sound/luther2.mp3|titles=luther 2]

FDR

[audio:/old/sound/fdr.mp3|titles=fdr]

JFK

LBJ

Richard Nixon

Gerald Ford

Jimmy Carter

Ronald Reagan

[audio:/old/sound/reagan.mp3|titles=reagan]

George H. W. Bush

[audio:/old/sound/hw.mp3|titles=h-dubya]

Bill Clinton

George W. Bush

[audio:/old/sound/bush.mp3|titles=bush]

Barack Obama

[audio:/old/sound/obama.mp3|titles=obama]

Code

OSC Record

I was amazed that no one had already done this, so I built a little program to record and re-send OSC (Open Sound Control) messages. You can download it here. Before making it run you’ll need to install the LIBLO library on your machine.

Music

New Tracks Summer 2011

I’ve made a couple new songs recently, both of which are a practice in film composition. The first is a study on the work of Eduard Artemyev, a Soviet composer who worked with the director Andrei Tarkovsky on the film “Stalker.” Here, I’m attempting to copy the affect of the films opening sequence, for which Artemyev wanted to create a mixture of eastern and western aesthetics.

[audio:http://amusesmile.com/old/sound/rasa.mp3|titles=Stalker Rasa]

From Wikipedia: “[Artemyev and Tarkovsky] finally found the solution in a theme that would create a state of inner calmness and inner satisfaction, or as Tarkovsky said “space frozen in a dynamic equilibrium.” Artemyev knew about a musical piece from Indian classical music where a prolonged and unchanged background tone is performed on a tambura. As this gave Artemyev the impression of frozen space, he used this inspiration and created a background tone on his synthesizer similar to the background tone performed on the tambura. The tar then improvised on the background sound, together with a flute as a European, Western instrument. To mask the obvious combination of European and Oriental instruments he passed the foreground music through the effect channels of his SYNTHI 100 synthesizer. These effects included modulating the sound of the flute and lowering the speed of the tar, so that what Artemyev called “the life of one string” could be heard. Tarkovsky was amazed by the result, especially liking the sound of the tar, and used the theme without any alterations in the film.”

The second is a track I made entirely from the sampled first measure of “In a Sentimental Mood” by Ellington and Coltrane. I thought of it as appropriate for a busy city street scene, but a close friend of mine says it sounds more like someone is loosing his/her mind. The beat starts at 0:36.

[audio:http://amusesmile.com/old/sound/July2011.mp3|titles=July 2011]
Code

Seattle Public Library Visualization

This is a visualization of transactions from the Seattle Public Library, taken from a dataset of over 66 million entries collected since 2005. Through this database, I was able to experiment with various methods to filter huge collections of information, attempting to extract meaningful patterns and depict them visually. Many thanks to Professor George Legrady for gaining access to this data.

In this version, I created a 3D map of the 11 most frequently referenced countries and the words most often associated with each for all of 2010. I did this by collecting the first three subject headings for every item circulated with a title referencing a country in either noun or adjective form (for example, “china” or “chinese”). This created a large text file which I simply parsed in terms of word frequency.

The result reveals interesting trends: “war” was the word most often associated with Germany, “cooking” and “food” were important categories for every country except England and Ireland (proving the fact that no one likes British food), and “women” was one of the most words most associated with France, etc. To add another layer of information, I’ve included scrolling “trackers” at the bottom that display actual titles of items containing these subject/keyword pairs (Italian + food, etc.).

The countries are listed in order of popularity- China being first- and the 3D blocks can be read as literal representations of the volume of items checked out in each category.

Technically speaking, data manipulation and filtering was done through MySQL in conjunction with some fancy BASH scripts, and all of the visualization was done in Processing. If you’re curious, please feel free to download the source code and try it for yourself.