AI noise-canceling headphone prototype unveiled: new scientific research results from the University of Washington

News on May 18th, a research team at the University of Washington recently made a breakthrough. They developed a real-time active noise reduction technology based on AI. This technology eliminates user-specified background noise without affecting the audio quality of the headphones.

This innovative team is led by Shyam Gollakota, a leader in the field of artificial intelligence. At a May 16 conference co-organized by the Acoustical Society of America and the Acoustical Society of Canada, Gorakota introduced the technology to attendees and demonstrated a working prototype live. This technology uses machine learning algorithms to convert human body movements into readable signals. By associating different actions with specific signal patterns, the team can capture and interpret limb movements in real time. This technology has a wide range of application potential and can be used for virtual reality interaction, remote gesture control, etc.

According to the editor’s understanding, Gorakota and his team used smartphones to The neural network system successfully identifies, trains and filters out 20 different environmental sounds. These sounds include alarms, alarm clocks, etc. that are common in daily life. This technology is expected to allow users to focus more on the sounds they want to listen to in complex sound environments and effectively block other interfering sounds.

Xi Lakota described a vivid application scenario in the demonstration: "Imagine that you are enjoying the sweet sounds of birdsong in the park, but the noisy people around you prevent you from fully immersing yourself in it. At this time, when you put on our noise-cancelling headphones, you can focus on the sound of birdsong, and those annoying noises will disappear. This is the original intention of our system design."

The prototype headphones shown. There are microphones on both sides of the earmuffs, connected to an Orange Pie circuit board via a USB cable. The board also provides audio signals to headphones via the audio jack. Judging from the configuration of the circuit board, it is likely to be the OrangePi 5B model, with a powerful Rockchip RK3588S SoC processor built-in. The processor has eight-core 64-bit computing capabilities and integrates a neural processor that can provide up to 6Tops of computing performance to achieve real-time noise reduction and filtering functions.

The team is proud to say that after training, their AI system is able to complete active noise reduction processing in less than one hundredth of a second without any dampening effect on the actual sound played by the headphones. .

The above is the detailed content of AI noise-canceling headphone prototype unveiled: new scientific research results from the University of Washington. For more information, please follow other related articles on the PHP Chinese website!