MGE specializes in the development of TWS headset charging bin, supports integrated circuit for conventional charging boost lithium battery protection three-in-one charging bin, and can develop charging bin with battery life display.
TWS Chip Solutions
Qualcomm IC/Realtek IC Chip Solution
50% reduction in power consumption compared with ordinary headphones;
Balance the power of both headphones for longer battery life;
Improved signal receiving and transmitting performance and stability;
Support aptX technology, minimizing the wireless audio transmission delay.
Low power consumption;
Support AI wake up;
Support Bluetooth 5.0, HFP1.7, HSP1.2, A2DP1.3, AVRCP1.6, SPP1.2, and PBAP1.0.
TWS Solutions Algorithm
Automatic Echo Cancellation (AEC)
When listening to music and TTS voice broadcast, it is essential to recognize wake-up words and command words. Meanwhile, echo is a crucial aspect that we have to consider. To minimize echo, we need to consider the structure and algorithm.
Algorithm support: priority should be given to the optimization of acoustic structure to reduce the echo intensity. On this basis, the echo cancellation algorithm can be used to further reduce the impact of echo.
Beamforming
Using the phase difference of the speech collected by the two microphones for processing to achieve directional pickup. This can enhance the voice of the user while suppressing the environmental interference and noise, achieving a high wake-up recognition rate even in noisy environment.
Algorithm support: Two microphones should be arranged one after another vertically with the microphone line pointing to the mouth. The distance between the two microphones should be 25 mm to 30 mm.
Noise Suppression
The beamforming algorithm already has a certain degree of environmental noise suppression, and noise reduction algorithm can further suppress the noise.
Algorithm support: With both beamforming and noise reduction algorithm, we can reduce the noise level by more than 20dB comprehensively.
Wakeup Word & Command Words Detection
In order to reduce the power consumption of the system and any invalid recognition, there is usually a custom wake-up word (such as Apple's "Hey Siri"). When the system detects wake-up words, it enters the process of speech recognition and interaction. In addition to wake-up words, there are some common commands related to music playing, answering phone calls, volume control, and etc. Since these commands are frequently used, ideally, users can directly say command words without first saying wake-up words.
Algorithm support: Currently, the algorithm can support the following Chinese and English quick command words, and can provide customized development for customers.