Mirelo releases SFX 2.0 video-to-sound-effect model
Berlin-based AI research team Mirelo released the video-to-sound-effect model SFX 2.0, which can modify a video's audio without regenerating the visuals, preserve dialogue, remove unspecified music, and generate sound effects, Foley and ambience synchronized with action and consistent across cuts.
Why it matters: The video-to-sound-effect model has been updated to redo the audio track without changing the visuals, and has been integrated into editing software and third-party platforms.