Neural network is able to voice the creations of artists
Neural network is able to voice the creations of artists

Neural network is able to voice the creations of artists

Нейросеть способна озвучивать творения художников (видео) A Japanese enthusiast by the name of Nao Tokui came up with a program based on a neural network that can select the most successful sounds and music to sound artists’ paintings. During the testing, the neural network coped with almost all the visual works, choosing the most correct sounds for them. . The neural network that formed the basis of the program is called Soundnet. She is able to determine what is depicted in the video, and then select the appropriate music to describe the picture. If, for example, you propose a neural network for video recognition with the image of dogs, the program will turn on the sound of a dog barking. In exactly the same way, the program works with images of children, adults, motorcycles, sports and music concerts, and other various objects. Nao Tokui used two sources to fill the program. From the Flickr service, he took various videos, and sounds from the free Freesound catalog. All files were passed by an enthusiast through Soundnet (this is the neural network in question). Thanks to Nao Tokui, a completely new and exciting application was invented for a non-new neural network.

Add comment