This research introduces HPM, a novel AI framework that leverages a latent diffusion model and a comprehensive film score dataset to automatically generate original and stylistically-controlled film scores from video input.
This research paper presents a series of improvements to Diff-A-Riff, a latent diffusion model for generating musical accompaniments, resulting in enhanced audio quality, diversity, inference speed, and text-driven control.
This paper introduces MT-MusicLDM, a novel multi-track music generation model based on latent diffusion, capable of generating coherent multi-track music, both conditionally and unconditionally, and excels at music arrangement generation.