KIMM researchers in South Korea have developed a prototype that extracts sound from the location of the person or object appearing on the screen. The system shared by the research institution implements the approach of producing sound by vibrating the screen in a way that changes the location of the sound source in real time. 80 vibration elements placed at the back activate selected areas instead of the entire panel.
Dongjoon Kim’s team focuses on the problem of image and sound coming from different directions. While the person speaking on a large screen stands at one end, a separate speaker can output the sound from another point. The prototype changes the region where the sound originates according to the movement of the source in the image.
There are voice coil motors behind the panel for this task. The system calculates the voltage to be applied to each element by analyzing how the panel responds to vibration. It then creates a vibration distribution that will ensure that only the targeted area emits sound.
In the team’s tests, target region selection accuracy reached 99.9 percent. The difference in vibration amplitude between active and inactive regions reached up to 35 decibels. These figures describe measurements in the research setup and are not a substitute for independent test results of a television on sale.
Researchers direct sound to selected area of panel
The approximately 100-inch prototype combines transparent LED film and aluminum panel. Using an acoustic camera, the team checked that the sound was coming from the planned points. He confirmed that when the source in the image moves, the area where the sound originates follows it.
In addition, artificial intelligence takes on the task of determining the location and size of sound sources in the image. The sound separation method selects the desired sound from the mix. The system reproduces this sound from the corresponding area on the screen.
The modular approach allows combining or separating parts according to screen size. Researchers are considering uses ranging from home televisions to exhibition displays to large LED panels. Working together of multiple modules also supports the production of low-frequency sounds.
The scientific article of the study appears in the Journal of Mechanical Science and Technology. The team also reports that it has received a South Korean patent for a related modular acoustic emission system. However, the institution does not announce a sales date or price for consumers.
In addition, researchers remove elements that cannot efficiently transfer vibration to the target area from the calculation. The method, which smoothes the transitions between consecutive audio signals, aims to reduce interruptions and unwanted noise.