Air transmission generates environmental sound such as human voices and noise through microphones. Bone conduction only collects audio information of vocal cord vibration of human (device wearer). The main function of the algorithm is to compare the sound signals obtained from these two paths to obtain speech and ambient sounds, and weaken the ambient sounds, ultimately leaving a relatively clean speech sound.

AI algorithm for noise reduction in single microphone calls
Implement TWS call noise reduction through software algorithms.

In terms of call noise reduction technology, in-depth learning is introduced into the field of signal processing, and AI call noise reduction and bone conduction sensors are effectively fused, which can achieve separation and extraction of voice and background noise, covering the whole scene with low signal to noise ratio. And this solution also only requires one microphone, which can improve engineering efficiency and save hardware costs, and is the first to achieve single microphone noise reduction.