Aesthetic investigations of synthetically produced SOUND COMPOSITIONS derived from variations in lou

NOTE: be warned synthetic sound created with computer may produce an uncomfortable sensation for your ears; please reduce the volume of your speaker and gradually increase it to match auditory perception without causing discomfort

In this work I explored the reduction of tango music to numerical data (loudness levels), and its re-interpretation into different sound textures that are detached from the original tango renditions. I was interested in investigating the possibilities of computers and software for the abstract and non-musical creation of synthetic sound compositions that could expand my own consciousness of what music is, and could be.

My creative approach used sound from tango songs as its primary compositional resource and textural material, with software as mediator of the creative action. In this sense, electronic sound emerged through 'playing' the software and thus, became an instrument for sonic exploration.

Because human action was primarily removed from the process of 'listening' tango and 're-interpreting' its musical data for its 'transcoding' into synthetic sound compositions, it pushed the boundaries of my perception and forced me to develop an extended awareness of sound that was in principle, deprived of any meaning. For this reason, the work presented here can be considered a computer controlled composition.

Sculpting sonic spaces

I approached the exploration of new sonic spaces by taking into account variations in loudness levels from tango compositions as textural material, reducing it to numerical data that was 'sculpted' into new sound frequencies by using an oscillator. The range of frequencies and the type of oscillator determined the texture of the sound produced but not its pattern. The pattern, or compositional aspect, was determined by variations of loudness levels in the original tango songs that were chosen. This means that I couldn't anticipate or envision the architecture of the piece beforehand because it resulted from the inherent data associated to each tango rendition.

Because computer controlled compositions were created by applying exactly the same algorithm to different set of renditions of the same tango song, architectural differences among synthetic compositions were solely determined by differences in tango renditions. However, the final pieces can be considered to be completely detached from the original tangos because their aesthetic outcomes didn't follow preconceived notions of human perception, they were the result of 'automated perception'.

'El Choclo'

In order to create different tracks of computer controlled compositions that would be sufficiently distinct from each other, an old and popular tango song was needed when considering diversity of orchestral renditions for appropriate heterogeneity of the textural material. Based on this, I found in 'El Choclo' an interesting tango song to experiment with. Composed in 1903 by Angel Villoldo and premiered during November of the same year by the orchestra of Jose Luis Roncallo, it has been rendered countless times ever since.

In this work, 24 different orchestral renditions of El Choclo were included from more than 40 I had access to:

Alfredo Marucci with Ramon Regueira

Anibal Troilo

Astor Piazzolla

Juan D'Arienzo

Carlos Di Sarli

Color Tango

Donato Racciatti

El Arranque

Forever Tango Orchestra

Giora Feidman

Juan Cambareri

Julian Plaza

Nuevo Quinteto Real

Olivia Molina

Puro Apronte

Susana Rinaldi

Florindo Sassone

Lalo Schifrin

Sexteto Canyengue

Sexteto Mayor

Sexteto Tango

Sinfonica de Montevideo

Tango 7

Tango Sur Trio

All renditions were used as textural material for the creation of 24 independent computer controlled compositions (herein CCC_1 to CCC_24) that conformed my first album on the topic. Each CCC retain the original length in minutes of the tango renditions that derived from.

Automated Perception & Interpretation

The computer algorithm I created played each tango song and extracted loudness levels as data values from 0 (no sound) to 0.5 (50% loudness) and mapped them to physical space along the y-axis of the canvas, as data points, in order to visually address the progression of the song and its variability in loudness levels (see previous essay here). Positional variation along the y-axis for each point was then translated into sound frequencies that ranged from 100 to 900, producing a synthetic sound through the use of a triangular wave with an oscillator object.

This approach resulted in the following sonic outcome:

Sonic Exploration 1. The first track corresponds to the orchestral rendition of El Choclo by Alfredo Marucci, and is the textural material that the algorithm worked with, extracting loudness levels through the song to create synthetic sound. The second track is the computer controlled composition (CCC 1) based on Marucci's renditions as result of automated perception and interpretation. This track is played with a frame rate of 1, and there is no variation in tempo across the composition. In the third track (CCC 1 1) variation in tempo was introduced by dynamically modulating the frameRate depending on loudness levels, giving the composition more of a playful feeling. Quality of sound recorded was low as the approach used involved me running one program that played the composition whereas a second program was recording it as sound emerged from the speakers of the same computer. By far this was not an ideal case, but sufficed to demonstrate proof-of-concept.

From this primary result, it was evident that controlled computer composition trascended the boundaries of what I perceive as tango, and even music. It did not delivered a pleasant experience, nonetheless it provided me with an expanded notion of the sonic space as well as the procedural ability to generate sound patterns based on rules applied to tango songs. The sound pattern of track 'CCC 1 1' greatly improved compared to track 'CCC 1' when tempo was introduced by exploring changes in frame rate across 'CCC 1 1' that were dependent on loudness level from Marucci's rendition. This prompted me to conserve this approach while extending the sonic space by using different oscillators objects (Sonic Exploration 2). Oscillators create distinct types of waves that produce different synthetic sounds.

Sonic Exploration 2. Track 'CCC 1 1' was explored by using four different types of oscillators: pulse (track 'CCC 1 2'), square (track 'CCC 1 3'), saw (track 'CCC 1 4') and sin (track 'CCC 1 5'), respectively. Only a few seconds were recorded for each piece as means to sample the sound characteristics for each oscillator.

I found sound textures given by oscillators producing triangle and square waves were the most appropriate for synthetic creation of sound compositions derived from tango renditions of 'El Choclo'. In the following experiment (Sonic Exploration 3), I played (and recorded) in parallel 'El Choclo' tango and its corresponding computer controlled composition with sound derived from a square wave. This allowed for cross comparison of human-machine perception in real time, and highlighted the tension between a very deep, emotional and meaningful tango composition as opposed to mechanic, synthetic, and electric rendition that was totally deprived from emotional meaning.

Album Prototype

Having developed the capacity for creation of algorithmically guided sound compositions prompted me to start experimenting with 'album prototypes'. This idea originated from a friend who suggested I could create an album with all these compositions. Although the recording capabilities that I have access to at the moment are not the most appropriate, the idea of album prototypes is an exciting one. I created the first album prototype by recording 10 computer controlled compositions derived from 10 different renditions of 'El Choclo' in which only the first half of each tango song was translated to computer controlled compositions. Thus, the tracks for the album draft are half the length of the original tango songs. Renditions included were as follow:

1 Troilo

2 D'Arienzo

3 Piazzolla

4 Di Sarli

5 Color Tango

6 El Arranque

7 Forever Tango

8 Feidman

9 Cambareri

10 Plaza

Additional tracks will be subsequently added from the original 24 renditions of El Choclo previously mentioned.

Is this Experimental Music?

According to what is written on the Wikipedia page referring to Experimental Music, it 'pushes the boundaries and genre definitions' and 'involves exploratory sensibility'. Although the computer controlled compositions presented here can be considered to push boundaries, and certainly involved exploratory sensibility, I am inclined to consider them as abstract explorations of sound that not necessarily need to be musical.


The author thanks Batt Johnson for suggesting I start thinking about creating an album; and Meg Farrell for suggesting I start considering tempo when working with sound.


A. Vandso (2016) Listening to listening machines: on contemporary sonic media critique. Leonardo Music Journal (26): pp. 48-52

Martin Knakkergaard (2016) Unsound sound: on the ontology of sound in the digital age. Leonardo Music Journal (26): pp. 64-67

Olav W. Bertelsen, Morten Breinbjerg, and Soren Pold (2009) Emerging materiality: reflections on creative use of software in electronic music composition. Leonardo (42): 3 pp. 197-202 (accessed on August 2017)

Philip Brody (2015) Sine qua son: considering the sine wave tone in electronic art. In Abstract Video: the moving image in contemporary art; edited by Gabrielle Jennings; published by University of California Press

Disclaimer: Original recordings of orchestral tango renditions uploaded to this essay as media is only for artistic and demonstration purposes.

Keywords: experimental sound; automated sound; synthetic sound; Argentine tango; New York tango; tango; computational art; computational abstract art; algorithmic art; Martin Calvino; sound sculpture; sonic spaces;