Odkrywaj magi臋 uczenia maszynowego w tej casualowej symulacji, kt贸ra zamienia ponad 44 000 pr贸bek audio z Freesound w nawigowaln膮 chmur臋 d藕wi臋k贸w 3D. Gracze przegl膮daj膮 osadzone w t-SNE dane zbioru, wybieraj膮 wyr贸偶niaj膮ce si臋 klipy i odtwarzaj膮 je, by tworzy膰 w艂asne instrumenty - jak pianina z syntezatorowych nut czy marimby z trzask贸w i p臋kaj膮cych d藕wi臋k贸w.
Proste mechaniki nap臋dzaj膮 niesko艅czon膮 kreatywno艣膰: przemierzaj rozleg艂膮 map臋 3D, zapisuj wybrane audio i uruchamiaj odtwarzanie dla rytmicznej zabawy. Ten indie eksperyment pokazuje, jak komputer radzi sobie z danymi d藕wi臋kowymi - idealny dla odkrywc贸w ceni膮cych eksploracj臋 ponad budowanie. Mi艂o艣nicy audio i casualowi majsterkowicze odnajd膮 tu plac zabaw dzi臋ki ogromnej skali i r贸偶norodno艣ci brzmie艅.
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
