Tauche ein in die Magie des Machine Learning in dieser entspannten Simulation, die über 44.000 Freesound-Audiosamples zu einer navigierbaren 3D-Soundcloud verwebt. Spieler durchstöbern die t-SNE-Embeddings des Datensatzes, fischen herausragende Clips heraus und spielen sie ab, um eigene Instrumente zu zaubern - etwa Klaviere aus Synth-Noten oder Marimbas aus Pops und Knistern.
Einfache Mechaniken entfachen endlose Kreativität: Bewege dich frei über die riesige 3D-Landkarte, wähle Sounds aus zum Speichern und löse Playbacks aus für rhythmisches Vergnügen. Dieses Indie-Experiment enthüllt den computerbasierten Zugriff auf Sounddaten und eignet sich ideal für Entdecker, die das pure Erkunden über das Bauen stellen. Audio-Fans und lockere Bastler entdecken hier einen Spielplatz voller immenser Größe und akustischer Vielfalt.
Feel the magic of Machine Learning in this new experiment that combines over 44,000 sounds from Freesound, a data set created by Kyle McDonald and a touch of Cabbibo weirdness.
Run your hands through a cloud of sound, learn the shape of the computers mind, select your favorite audio delicacies and make a drum set from this decadent aural smorgasbord.
The mechanics may be simple ( navigate through a 3D map of sounds , select individual sounds to keep, and hit those sounds again to play them ), but because of absolutely massive size of the data set, the possibilities are endless!
Want to create your own piano out of synth notes? Go ahead!
How about a marimba of pops and crackles? Be my guest!
And if you are just the exploring type, the data is beautiful enough to begin with! There is no need to make an instrument, or play it, because in from of you lies a tool for discovering and understanding a beautiful algorithm, and even more beautiful data: Sound
This project would not be possible without the amazing community of FreeSound.org
As well as Kyle McDonald, who scraped audio from freesound.org & made t-SNE embeddings of the data!
Created during Unity Artist in Residency 2017
