Turn abandoned patents
into your own filing
into your own filing
This invention describes a smart computer program, a neural network, that listens to sounds through several microphones. It can identify sounds, make them clearer, and even figure out where they're coming from and how far away they are, all without needing extra equipment. The system learns by analyzing sound recordings made in a special echo-free room, in computer-modeled environments with different echoes, and in real-world settings, with the claims specifically detailing the echo-free room recordings are made by rotating the microphones.
Why it matters: Deep learning architectures and computational power for training complex neural networks have advanced significantly since 2016. This makes the direct learning of spatial audio features, replacing traditional beamformers, more practical and performant.
AI gives you a few directions you could take this. Pick one, and we check whether your version is different enough to patent, then write the filing.
Reinvent this with AI
