Neural network generated sound and imagery (2021)
An experience of tranquillity often includes environments with natural features and the absence of human activity. How would a machine interpret this feeling?
tranqGAN A.I. is a deep neural network that is trained to hallucinate upon this concept and generate both audio and images related to this.
To generate images the system is trained using a styleGAN2 network and images available online. The audio is synthesized using a sampleRNN network and have been trained on audio recording of birds, water and wind.