Noise-canceling headphones are perfect for those who want to eliminate all the clutter, but often hearing the right thing at the right time can be important.
However, people can’t choose which sounds their headphones can cancel out.
3
3
This can lead to potentially dangerous situations, such as a pedestrian not being able to hear a car’s horns while walking near a busy street.
Researchers at the University of Washington have developed deep learning algorithms that allow users to choose which sounds to filter through their headphones in real time.
The team calls the system “semantic listening.”
The headphones capture audio for a connected smartphone, canceling out all environmental noise.
Through voice command or an app, users can choose which sounds they want to include from 20 categories, such as sirens, baby crying, speech and bird chirping.
Only the chosen sounds will be played through the headphones.
A video showing examples of the system in use was uploaded to YouTube and has over 3,760 views at the time of writing.
The team published your discoveries and there is future plans to launch a commercial version of the system.
“Understanding the sound of a bird and extracting it from all the other sounds in an environment requires real-time intelligence that today’s noise-cancelling headphones have not achieved,” Shyam Gollakota, professor at the Paul G. Allen School of Computer Science and Computer Science. Engineering, said UW News.
“The challenge is that the sounds that headphone users hear need to be synchronized with their visual senses. You can’t hear someone’s voice two seconds after they talk to you. This means that neural algorithms must process sounds in less than a hundredth of a second.”
Time crunch forced the team to have the semantic auditory system process sound on a device like a connected smartphone rather than serving it in the cloud.
Additionally, because sounds from different directions arrive at a person’s ear at different times, the system needs to preserve these delays and other spatial cues.
This is so that a person can still perceive sounds in their environment.
The system was tested in various environments such as offices, streets and parks.
What is semantic listening?
Semantic hearing is a new capability of hearing devices that can focus on or ignore specific sounds in real time.
The system can also preserve spatial cues in real-world environments to prevent users from potential danger.
This could benefit people who need focused listening for their jobs, such as healthcare, military, and engineers.
It can also help factory and construction workers who want to protect their hearing.
In these tests, the system was able to highlight sirens, bird calls, alarms and other noises, while removing all other real-world sounds.
Around 22 participants evaluated the system’s audio output for the target sound, saying that the sound quality has improved compared to the original recording.
However, in other cases, the system had trouble distinguishing between sounds that share many properties, such as vocal music and human speech.
The researchers said training the system using more real-world data could help with these results.
This story originally appeared on The-sun.com read the full story
