1 paper
William Zerong Wang, Dongfang Zhao
Modern music retrieval runs on vector embeddings, and once these embeddings are shared for search or matching they can be copied, probed, or used to train generative models. Fully…