David Michael

 A Deep Forest Creek


A Deep Forest Creek is a 19 hour 36 minute continuous audio recording taken in 2 stereo channels near Lake Mamori in Amazonas, Brasil. The recording was released as a DVD-ROM and as a single memento: a standalone player with outputs for headphone listening or connection to loudspeakers.

The recording was made as during a residency at the Mamori Sound Project organized by Francisco Lopez from November 4th – 17th, 2009. The memento was produced between December 2009 and January 2010 in Tarrytown, New York.

The thickness of the present

The area around Mamori Lake, in the Brasilian state of Amazonas, is classified by the World Wildlife Fund as tropical moist broadleaf forest. More specifically, it is part of a terrestrial ecoregion characterized by flooded forests that are seasonally inundated by overflow from the rivers of the lower Amazon [6]. Stable temperatures, consistent light, and copious amounts of water drive a relentless turnover of organic matter that supports extremely high biodiversity.

The sound environment of the tropical forest is incredibly dynamic. The actuation of millions of living things interact in specific spaces and times to produce collective sound structures at many different time scales, frequencies, and intensities. Some of the temporal structures, like the chorusing of frogs, are relatively easy to hear. But the structures at much longer durations, like the sound of day, are somewhat outside our conscious perception. We can describe in words how the sound environment of the forest evolves over the course of the day, but it is incredibly challenging to be engaged in the act of listening for this period of time.

At long timescales, our experience is perceived through memory. The “thickness” of the present moment is actually quite minimal, and is intimately tied to our physiology. The pressures of metabolism, the environment, and the temporal range of our own neural oscillations limit our ability to experience a day as a complete structural unit [2,3,5]. We attend to shorter, more local events one at a time and then reflect on the entirety of what has happened.

Memory devices

To assist us in the experience of time, external memory devices are used. The use of things or techniques that assist memory is probably as old as our humanness. At their most basic, memory devices can be physical artifacts that relate us to an experience. We even use our interaction with the environment itself as a form of memory, such as a broken branch or a trail marking our path through the forest (and potentially the way home). In modern times, technologies from writing to electronics have vastly expanded how we might “remember”.

The two memory devices I am most interested in, mementos and recordings, facilitate memory in very different ways. Mementos are kept to remind their possessor of a person, place, or experience. They are not typically seen as a representation of the thing they reference, but rather only as spark for memory and thought. Even in a technological society, we actively collect mementos to help us later reflect on entire portions of elapsed time. A stone, a feather, a pressed flower.

At the opposite extreme, we use recordings to very literally memorize events captured from the environment and even from within ourselves. Current electronics technologies allow for the “fixing” of at least some sort of representation of natural events at nearly arbitrary durations. That is to say, devices that can record sound continuously exist [4], though the interpretation of these recordings by either man or machine must be challenging. Through recording, we supplement our memory with the ability to revisit specific sequences of events indefinitely.

To reveal the sound structures of a day in the forests around Mamori, it is helpful to offload the task of memory to electronic means. But a recording of a day no more assists our perception of a day as a structural unit than if we had been able to sit at the location in the act of listening. While a recording retains some aspect of the forest for review, it really is something else entirely. The recording completely mediates the experience of the forest, that is, the recording is the experience [4].

The fact is that mediation is required to experience much of our natural world, and this is particularly true with natural sound. It is not just the timescales on which natural events unfold that make mediation necessary, often the environments in which natural events take place are inaccessible, inhospitable, or the inhabitants are simply wary of human presence.

The recording and its embodiment

A Deep Forest Creek is a 19 hour 36 minute continuous audio recording taken in 2 stereo channels near Mamori Lake in Amazonas, Brasil. The recording is housed and played back from a wooden box (a memento), with outputs for headphone listening or connection to loudspeakers. Playback begins when the box is turned on, and ends with the completion of the recording, or when the box is turned off. The exact duration of the recording reflects an attempt at capturing a full 24 hours of the sound environment, limited by the logistical factors of deployment and retrieval as well as the weather.

The recording itself was made as part of the Mamori Sound Project from November 4th – 17th, 2009. Mamori Sound Project is an artist residency and workshop organized by Francisco Lopez and held at the Mamori Art Lab located on Mamori Lake. The box which contains the recording was produced between December 2009 and January 2010 in Tarrytown, New York.

While the events recorded at the creek were overwhelmingly public (though without human audience), the presentation as a memento suggests to a listener that the experience of A Deep Forest Creek is a private one. This is perhaps true of all mementos. There are two outputs for headphones so that the memento can be shared, but only so much. Each output contains the stereo field of a separate set of microphones. Many of the events happen simultaneously in both headphones, but the stereo fields are are different.

A Deep Forest Creek reflects two forms of memory devices. The recording is a device in the technological sense, fixing and playing back events from an electromagnetic memory, in this case a CompactFlash EEPROM. The embodiment of the recording suggests a memento, which is a memory device in the sense that “device” refers to a technique – a memory helper. While traditionally mementos evoke memories of past experience, here the memento itself contains the entire experience. A tiny world captured in a box.

Its specific embodiment also reflects a desire to present the recording in a medium that is more persistent, personal, and accessible than a multichannel digital file. It should not be up to the audience to provision software or hardware for playback. The medium not only stores the content, but is also capable of playback. It is somewhat insulated from the shifting formats of consumer audio. Today’s multichannel FLAC is yesterday’s quadraphonic vinyl.

Program and setting

The creek itself is located a few hundred meters into the forest on the South bank of Mamori Lake. The creek is part of a truly vast network of waterways that drain the surrounding forest. November is the dry season in the Brasilian Amazon, and during this time of year the creek’s water has slowed to a trickle, trapping fish who can only wait for rain. The banks of the creek slope upward for about 10 meters on either side into the dense forest. With so little rain, the creek is a destination for local wildlife. Some of the vocalizing animals can be identified, many cannot.

The recording begins close to 4pm, just before nightfall. As insects come alive in the cooling forest, Screaming Piha (Lipaugus vociferaras) fill the early evening air, joined with bursts from a Red-billed Toucan (Ranzphastos tucanus). Soon after sunset, frogs and toads call to one another from the creek. Small nocturnal mammals rustle in the leaf litter. The calls of insects dominate the night, punctuated by the social calls of bats. Past midnight and into the early morning hours, Howler Monkeys, presumably Alouatta seniculus, display to one another. Dawn brings a small explosion of birdsong which slowly gives way again to the Screaming Piha. As midday approaches, a thunderstorm gathers with peels of thunder.


Two pairs of microphones were used at the site. One pair of omnidirectional Audio Technica AT-4022's mounted in a parallel boundary array and placed near the edge of the water. About 9 meters up the creek bank, a Sennheiser MKH 30/40 MS array was pointed in the opposite direction into the forest. Both microphone pairs were mounted on tripods about a meter high and covered with small umbrellas to protect against unexpected precipitation. The microphones were connected to an Edirol R-44 multichannel recorder.

For the construction of the memento, a small embedded computer running Linux was used to control the playback over 4 channels. Two stereo 1/8" phono inputs are provided for headphone listening. The innermost input is the front stereo pair (AT4022) facing creekward. The outermost input is the rear stereo pair (MKH 30/40) which faces into the forest. Additionally, there are indicator LEDs for power and disc activity. A single button turns the machine on and off.

Audio was recorded to 16 bit/48 KHz BWF files. Equalization and limiting were applied to each stereo recording independently. The final file was encoded to multichannel Ogg Vorbis at 48 KHz/640 kbps CBR.


Thanks to Slavek Kwi and Francisco Lopez for suggested the location and guiding me on the deployments as well as everyone involved with the Mamori Art Lab. Els Viaene was incredibly kind to lend her MKH 30/40 pair for deployment after mine began to sputter in the humidity (I know, impossible right?). Thanks also to Rob Danielson for his technical advice and experimental push.


1. Echelon (signals intelligence).

2. Gyorgy Buzsaki. Rhythms of the Brain. Oxford University Press, 2006.

3. A. P. Levich, editor. On the Way to Understanding the Time Phenomenon. World Scientific Publishing Company, 1995.

4. Franscisco Lopez. Environmental sound matter. 1998.

5. Curtis Roads. Microsound. MIT Press, 2002.

6. Robin Sears. Monte alegre varzea (nt0141). Technical report, World Wildlife Fund, 2001.