2023-12-25 19:30:00
How about having one headset that can only block some specific types of sounds? That’s why researchers at the University of Washington have developed a system capable of filtering some of the audio in the environment, thanks to the use of artificial intelligence, promising a personalized experience.
The new feature was called “semantic listening” and aims to enable people to isolate only unwanted noise. For example, like sirens or babies crying, however, keeping the hearing of some more pleasant ones, like birdsong. The system works with deep learning algorithms.
These algorithms are responsible for filtering sounds according to the user’s preferences. Furthermore, the equipment makes use of a neural network running on a connected cell phone. This way, noises are sent to the device, which blocks unwanted noise. In fact, it is possible to select audio from 20 different categories to play.
Bandhav Veluri is a PhD student in Computer Science and commented on the news:
The target sounds we play through the speakers would have to be consistent with what the user sees in the environment. For this reason, we only have 20 milliseconds to process the incoming sound, extract the target sounds, and play them back to the headphones.
Even though it operated well in tests, the system is still not 100% efficient and still has difficulty differentiating sounds such as singing and speaking. Therefore, the tendency is for it to be improved in the future, which should include training with real-world data. Still, the developers believe that it is ready to be launched on the consumer market.
That said, it’s possible that we’ll see headphones released with more complex and intelligent noise-canceling functions in the future. Speaking of which, check out our guides to help you buy the best on-ear headphones and the best in-ear models available on the current market.
1703555278
#Researchers #create #headphones #selectively #cancel #noise