If you are young, you have no idea how to identify a particular voice among other voices in a crowded hall or a factory. But as one ages, one’s hearing capacity may reach a stage where a noisy hall becomes just a jumble of noise.
AI Based Headphones
The University of Washington researchers have recently introduced what they refer to as Target Speech Hearing. It is an AI-based headphones that allows you to stare at a person and isolate their voice from the background noise.
However, for the best results, one has to enrol their voice first and that is why it would not make a good eavesdropping tool.
If you want to know more about how it works, they explain it in their paper. The prototype employs a Sony noise cancelling headset. However, it uses binaural microphones, which means two extra microphones are attached on the outside surface of the headphones.
If we have the training data, then we can ask ourselves whether the traditional methods of correlation would also be effective.
In other words, you could use facial recognition to determine who is speaking and then filter out the voice using signal processing.
However, this system can potentially pick up sound from unknown speakers, figuring direction from the binaural microphones, so even if the correlation method worked well on known speakers, the new system is likely superior in new situations.
As you will see, there is more to noise cancelling headgear than first meets the eye. Or you can just go low-tech.
Also read: Progression of Artificial Intelligence: How Advanced is AI Right Now?
