When trying to hear what another person is saying, hearing aid users are often stymied by loud background noises, such as the voices of other people in the same room. Scientists at China's Xi'an Jiaotong-Liverpool University are working on a solution, in the form of a "smart" camera-connected hearing aid.
Currently-available hearing aids already often use noise-cancelling algorithms to filter out unwanted audio frequencies, allowing speakers' voices to be better heard. The proposed "cognitive hearing aid" would take things further. Being developed by a team led by Dr. Andrew Abel, it would use a linked wearable camera to filter out problematic noise much more effectively.
By imaging a speaker's mouth, it is hoped that such a system would be able to predict the vocal sounds that the person was about to make, then respond by boosting the associated frequencies while dropping others. This is somewhat how our brains work, as we combine what we hear the person saying with what we see their mouth doing.
To that end, Abel's team has already created an algorithm that can track the movement of a person's mouth, detecting whether it is open or closed, and gauging its width and depth when open. It works on a wide variety of people, and requires minimal processing power.
This system could conceivably even be used for lip-reading, filtering out background noise based on the actual words that the person is speaking. Additionally, the technology may be able to visually recognize different environments – such as quiet offices or noisy bars – and apply noise filters specifically suited to such places.
"When we talk to each other, we don't just rely on sound," says Abel. "We look at each other's faces, we look at each other's body language, and we all lip-read to an extent. So far, we've been unable to incorporate these things into hearing aid technology. That's ultimately what we're looking to change."
