DiffRhythm

DiffRhythm AI is a song generation model based on latent diffusion technology. By simply inputting lyrics and style prompts, it can quickly generate complete songs with vocals and accompaniment, with a length of up to 4 minutes. It supports a variety of music styles, and users can control the music style through text prompts to generate music in different styles ranging from rock to jazz. Its non-autoregressive structure ensures rapid generation speed, and the architecture is scalable, enabling it to be trained on large-scale datasets to continuously improve and expand its capabilities.