A new approach for chat sonification

This paper presents a new approach of chat sonification based on a deep neural network for music generation. The advantage of chat sonification is that the feeling of something artificial in the generated music is less present than for other artificial intelligence approaches. The new method of sonification introduced in the paper is based on the polyphonic model theory, similarly to the idea used in the MusicXML Creator Platform, the novelty being that it uses sequence-to-sequence neural networks for an improved pitch generation. The duration generation algorithm remains the same as it was implemented in the MusicXML Creator. The results are evaluated by multiple participants