Skip to main content

Dubnov Shlomo speaking in a conference room. The logo for the International Conference on Acoustics, Speech, and Signal Processing.

 

Two research papers by graduate students Tornike Kharkchadze and Keten Shao with Prof. Dubnov will be presented in the ICASSP conference in April 2025. Below are his research papers:

  1. Simultaneous Music Separation and Generation Using Multi-Track Latent Diffusion Models 
  2. kNN-SVC: Robust Zero-Shot Singing Voice Conversion with Additive Synthesis and Concatenation Smoothness Optimization
These works present new generative AI methods for music and voice. The first paper deals with automatic generation of music across multiple tracks and the second with substitution or invention of new artist voices in a singing recording. These papers present SOTA technical innovation in a major engineering conference in the field of acoustic signal processing, and are expected to open novel creative possibilities for music composers and producers.