Noise reduction (active noise reduction + call noise reduction)

Active noise reduction is mainly used to play audio, monitor the environmental noise through headphones, and then emit sound waves with opposite phases and equal intensity. After the two are superimposed, the noise is canceled. To achieve such a function, the headset needs an outgoing microphone to receive ambient noise, and the signal processing requires a certain time. Therefore, the main control chip needs to make a prediction of the future noise through a certain algorithm, and then play when the noise comes An opposite sound wave cancels it. Some high-end products also have an inward microphone, which is used to monitor the actual playback sound and provide feedback to achieve further calibration and optimization. At present, the latest TWS chips of Apple, Luoda, Hengxuan, Qualcomm, Realtek and other manufacturers have already supported the active noise reduction function, which is also the starting point for various manufacturers to improve the user experience

In addition to achieving noise reduction when playing audio, call noise reduction is also essential. The purpose of call noise reduction is to selectively input the user's voice and filter the ambient noise when the user is speaking, so that the other party of the call can hear the user's speech more clearly. However, compared to active noise reduction, call noise reduction is more popular. At present, the mainstream call noise reduction technology on the market is a dual-mic noise reduction scheme. The principle is to capture voice signals through two microphones and filter other ambient sounds.

Generally speaking, the talk microphone at the end of the TWS earphone handle is closer to the mouth, so the received voice will be much louder than that of the outgoing microphone, and the source of the ambient noise is about the same distance from the two microphones. The environmental noise intensity received by the two microphones is almost equal, and the differential amplifier between the two microphones subtracts and amplifies the signals received by the two microphones, and the environmental noise with similar reception intensity is filtered out, leaving behind The voice with a large intensity gap ensures the clarity of the call. Some high-end products also further detect the vibration of human muscles and bones through the internal voice acceleration sensor to determine when the person is talking, so as to achieve a better sound pickup effect.

The other solution is AI call noise reduction technology. In 2018, a single-microphone call noise reduction solution proposed by Elephant Sound Co., Ltd. combines computational auditory scene analysis theory (CASA) with deep learning technology to separate people in real time. Sound and background noise, extract clear human voice from environmental noise. OPPO's TWS headphone product OPPO Enco Free uses the sound plus technology's SVE AI dual wheat noise reduction technology, which combines the two, combining deep neural network with dual microphone beamforming technology to achieve a considerable noise reduction effect .

Principle of double wheat noise reduction

Post time: 2020-09-01