Convex Audio's patented technology is powered by a new class of algorithms, which pushes the state of the art in multi-microphone processing. These algorithms share some features with commonly used techniques, including superdirectional beamforming and blind source separation. However, they are based on a modern, more recent statistical framework, which leads to important and favorable differences. One leverage this framework has, for example, is the ability to incorporate features of human speech.
Using the new framework helps the algorithms separate voice from ambient sounds more effectively, and better adapt to properties of the noise and the acoustic environment. This often results in a cleaner and more accurate voice reconstruction than existing techniques.