AIELSON: The Time of Man

My residency (part of Google's Artists + Machine Intelligence program) consisted of studying the Quipu and reviewing the possibilities of Machine Learning for furthering its comprehension. I wanted Quipus to talk again. But, for that, I needed a voice, but not any voice: one that would come from beyond life, from the afterlife that audiovisual media facilitates when it allows the leaving of sound or video recordings of people that already passed away for posterity, in a way living forever embedded in memory, while belonging to another history.

In the text included below, Eielson speaks of a “doll that will be fed with the most important poetry in the world” and that “will continually recite through the doll's mouth the most beautiful poems conceived by man.”

Using a Generative pre-trained model retrained to Spanish poetry and finetuned to the poetry of Jorge Eduardo Eielson, Cecilia Vicuña, and myself, yet first the model was trained with large amouns of poetry in Spanish.

I was intrigued by the possibility of picking up the stylistic qualities of speech that could be recognized as characteristic of the South American region. I chose the data for the finetunning based on common things among the artists and myself: as sharing an Andean origin, an interest in visual arts, sound, and poetry, but, most importantly, the appreciation for Quipus, referencing, as well, an alternative history of art and technology. As I was investigating voice and accent, a deep Learning system consisting of the Tacotron 2 + Waveglow speech synthesis pair was trained, first, with Argentinean speech + text, and then with the voice of Jorge Eduardo Eielson, in a way, bringing him back from the dead.

The generated voice is not particularly modulated nor clean, unlike what we are used to by now as synthetic voices tend to pronounce understandable phrases: not all the words are intelligible, something that, for our ears trained in the linguistic perfection of the machines, feels familiar and human. This sound art poems are not intended to be passed as if they were his, on the contrary –this is presented as a third entity, “Aielson” that does speak like an old Peruvian man with a Peruvian accent (although sometimes dilucidating its originally Argentinean background) in a way, fusioning Latin America in a male voice.

The story of the Voyager 1 probe is known, sending an easy-to-play disc into space in 1977, with a recording of human voices greeting in 55 different languages, targed to a space and / or future culture. I here I do something similar: wanting to move away from the digital, I decided (at this point, already fully aware of the connection with the “Horrifying Sculpture” and working based on this) to go back to an analog format, that is, vinyl: to turn electric impulses onto grooves grazed into a plastic surface so, like the hand that once counted the Quipu's threads and knots by contact with it, the needle will brush the curves of the vinyl to reproduce sound. Just as ceramics was once the main material used by inhabitants of these lands, nowadays, undoubtedly, is plastic: we can imagine future intelligent life forms finding abundance of plastic "huacos" from our time, one of this being the album entitled “El Tiempo Del Hombre” (In the Time of Men, one of the verses that the Aielson system came up with). Having an analog support will allow a rather easy reproduction of its contents, which will appear as an old man from the early 21st century reciting poetry in Spanish.