At present, there is a mature sound localization algorithm in the industry, which is called Head Related Transfer Functions (HRTF). It mainly uses the algorithm to turn a single sound source into two channels to deceive people. The specific principle is explained and explained in detail in wiki or elsewhere. Simply put, after a sound is emitted from a certain position in space, it will be introduced into two ears and reflected by the auricle. And finally reach the eardrum. There will be a time difference between the arrival of these two sounds, which is a major factor in people's positioning. As for the specific directions such as up and down, back and forth, it mainly reflects and filters the sound through auricle, external auditory canal and many other places. This discrimination is highly personalized. Everyone's ear canal, auricle and even skull are different in size and shape, and everyone's ability to locate sound has been trained for a period of time. For example, if the auricle is injured and cut off, people's ability to locate sound will be damaged, but after a period of adaptation, it can still return to normal.