Panasonic Holdings Co., Ltd. (Panasonic HD) and Panasonic R&D Company of America (PRDCA), in collaboration with UCLA researchers, have developed OmniFlow, a multimodal generative AI capable of converting between text, images, and audio ("Any-to-Any").
OmniFlow reduces training data costs by combining specialized generative AI models (text-to-audio, text-to-image) and learning high-precision models with minimal multimodal data. This breakthrough was presented at CVPR 2025.
