Fire is one of the key elements in Nature, used from the prehistoric era to heat living environments, to cook food for survival and enjoyment.
Challenging the silent nature of fires, we utilize an assembly of deep neural networks to find a way to better understand the underlying mechanics of fire, and render complex but silent flames audible by associating them with a unique sonic spectrum.
The total image of the fireplace is segmented into small partials, each of which is analyzed for its sonic equivalence in terms of the local flame features, then rendered into audible signal using wavetable synthesis.
This offers a thorough exploration of the deep, complex structural content of the flames in this fireplace.
Source by Markus J. Buehler
Add comment