Imagine being at a party, trying to listen to one person while tuning out everyone else around you. Humans are naturally good at this, but technology has struggled with the same task, which is called the “cocktail party problem.” However, thanks to AI, there’s now a solution.
Engineer Keith McElveen, who worked on a war crimes case for the U.S. government, became interested in solving this problem. His company, Wave Sciences, developed an AI that can separate overlapping voices by analyzing where the sounds in a room come from. This technology is a game-changer in court cases, where background noise can make it hard to use audio evidence.
Wave Sciences’ technology was recently used in a U.S. murder case, where it played a key role in the convictions by filtering out unwanted noise and focusing on important voices. The FBI needed clear audio from a meeting in a noisy restaurant, and this AI technology made it possible. Other labs, including in the UK, have tested it too.
This tech could also help in situations like hostage negotiations, where it’s crucial to hear both sides of a conversation. McElveen’s AI is being marketed to the military and could even be used in smart speakers and cars, helping these devices hear commands clearly, no matter how noisy the environment.
AI is making big strides in forensics, with machines now able to analyze voices, detect alterations in recordings, and ensure the integrity of evidence presented in court.
Companies like Bosch are also using audio AI to predict machine malfunctions before they happen.
Interestingly, tests have shown that this AI algorithm works in a way that’s very similar to how our human brains process sound.
McElveen believes that in solving the “cocktail party problem,” they might have uncovered how our brains work when we focus on specific sounds in a noisy room.
According to the BBC, this discovery could have a major impact on both technology and our understanding of human hearing.
This new technology could soon be everywhere, making it easier for devices to hear and understand us, even in the noisiest situations.
Credit : BBC
https://www.bbc.com/news/articles/c5yk5mdj9gxo