Tag
Released a model (Stereo2Spatial) that converts stereo music tracks to spatialized binaural mixes, using flow-matching diffusion and amplitude lifting for stable training. The model and a Windows app are open-sourced under Apache 2.0.