Wait until you see the same concept combined with NeRF idea. The output won’t be 3d shapes but another model that can generate realistic and geometrically consistent images of a scene viewed from different angles.
Patent what? Supervised sound localization is not novel. There exist myriads of published work on that topic. The only novelty I see in this work is the similarity between performance of their trained model and that in humans.
This neural net could create a realistic 3D environment where AI agents could train and explore. The lack of embodiment is the main thing missing from reinforcement learning agents to close the gap to humans. The environment itself can be controllable by embeddings, spanning any conceivable situation. Of course, ideally, if this architecture could be made much more efficient.
Maybe you could use a tamper-evident security token along with the password. Could be as simple as a fortune cookie. Keep it safe and crack it open at the end of the year to write the fortune next to your password. In the case of a probable time traveler who tells you your password, challenge them to tell you your fortune and open it afterwards to verify.
