back home? 🌻

art

i try to do all kinds of things! so for your troubles, here’s a table of contents:

improvised music
with neural nets
web sketches
live coding
electronic music
tunes

improvised music

live at elastic arts | improvised music series (2024)

stretchmetal presents | drone rodeo (2024)

the air is full (2024)

with Tyler Felson

sound collage - march 28th, 2024

dueto - mi má y mi pá (2024)

introspections (2023)

Cccccccc (2023)

Working on a patch for Francisco Corthey. Granular looping.

sunday (2023)

hugo flores-garcía · sunday

with neural nets

token telephone (2024)

A neural sound installation by Hugo Flores García and Stephan Moore.

at the NIME conference in Utrecht, NL, Sept. 4-6, 2024.

Token Telephone is a co-creative AI sound installation where participants interact with a chain of generative AI models, initiating a generative game of telephone. The installation space is circled by four neural networks, each represented by a loudspeaker. Participants make sounds into a microphone at the entrance of the installation space. Their sounds are iteratively transformed by each neural network in a feedback loop, deviating further from the original with every pass. This iterative process reveals patterns between the input and the training data of the networks, morphing human utterances into new and unexpected sound textures.

world of mouth (2024)

world of mouth is a fixed media 8-channel composition for a generative sound neural network and voice.

In world of mouth, I construct several sonic environments entirely out of gestures recorded with my own voice. These vocal gestures are used as guidance for a generative model trained on a sound library of percussion instruments and natural sounds.

The vocal gestures gave timbral and rhythmic shape to the generated sounds, which contain the sonic properties of the collected material (e.g. percussion + natural sounds), while still following the patterns of my voice. The result is a fascinating morphing process between sound textures: we can listen my vocal gesture give shape to the sound material, but eventually, the sound material bites back, imposing its own rhythmic properties into the resulting sound.

If you would like a copy of the 8-channel version, email me hugggofloresgarcia [at] gmail [dot] com.

chicago creative machines (2024)

I had the joy of giving the inaugural talk + performance for the Chicago Creative Machines series at ESS Chicago on Feb. 25, 2024. I talked about my compositional work with vampnet, using the mouth as the interface for a generative model, showed an 8ch composition for voice and vampnet, and played a solo set of instrumental music with.

confluyo yo, el ambiente me sigue (2023)

confluyo yo explores the raw, timbral and rhythmic relationships between intentional sounds, natural sounds, and sounds as byproduct of human activity. in the piece, a horn player (Michael J. Pineda) plays short 9 second themes called ‘seeds’ and improvises over them. the seeds are repeated for the rest of the piece, though, through repetition, a generative model slowly “dematerializes” or transfers the timbral characteristics of the saxophone to the sound of birds, or the sound of machines.

Read the program notes and score here. confluyo yo makes use of VampNet. happy to announce that confluyo yo has been accepted for the ISMIR 2023 conference and will be performed in Milan, Italy in November 2023!

Recordings available on Bandcamp.

salad bowl (2023)

salad bowl is a neural sound installation that invites participants to co-create a sound collage with the salad — a neural network trained on different distributions of sounds.

to begin the interaction, a participant must make a sound into the microphone in front of them. the neural network then takes the participant’s sound as input and destroys around 80-90% of it, then attempts to regenerate the original sound — though it knows very little about the original sound (human speech)! however, it does know a lot about natural sounds and specific musical styles. this process is repeated a number of times in a feedback loop and shown to the participant.

— the end result is a fascinating auditory look into the processes behind generative sound, as well as the ability of neural networks to spot and leverage relationships between sound objects in even the most unintuitive of circumstances.

living // dreaming (nightmare 1) (2023)

Exploration of the resonances in a partially-trained raw-audio AI model trained on pop/lofi/breakcore/hip-hop music. High entropy, short structure, in short loops.