Wearables have solidified their position in the global mobile market in the form of smart watches and smart bands. Today, a new form of wearable devices, called Earable, is emerging. Earables are essentially wearables worn in the ear. With the wide adoption of earbuds, there is a great opportunity in using them for 3D sound spatialization, or namely Acoustic AR. Its use cases range from recreation activities like video games, to health sensing and education scenarios. On the other hand, Earables can pose safety risks to users wearing them all the time. Noise cancelation in earbuds isolates users from surrounding ambient sounds, making them unaware of any imminent dangers such as oncoming traffic or attackers. There is a need for an artificial auditory system that can localize and detect relevant sound events.

Earable is an embedded system of different sensors, which are packed within an earbud for different quantifications. Due to its small form factor and wireless transmission capabilities, Earable will provide users with much more comfortability while doing their daily activities. They are the most suitable form of wearables to deal with the biosignals of the head because of its location. Moreover, it can use proprietary machine learning algorithms for higher accuracy of the bioelectric analysis. Basically, it is an earbud where the user can use it for common applications like listening to music, or answering a phone call, as well as acoustic AR like video games and education, while in the background, different sensors are continuously monitoring the biological signals of the user and use them for health monitoring and other scenarios.

Humans can hear and localize the incoming sound to the ears at the same time. This ability is attributed to filtering effects of the head and body on the incoming sound, which is called Head-Related Transfer Function (HRTF). HRTFs are frequency and direction dependent. Moreover, since the physical properties of different people are not the same, HRTFs are also subject specific, which makes it hard to estimate. Learning the HRTF of each human subject makes it feasible to do sound externalization (or spatialization), as if the HRTF of another person or a general one is used, the sound does not appear to be natural, and leads to bad user experience.