One of the key challenges in acoustics, known as the “cocktail party problem”, has finally been solved thanks to artificial intelligence (AI). This problem relates to the ability to distinguish a specific voice in a noisy environment, which has been very difficult to achieve with technology until now.
Keith McElveen, founder of Wave Sciences, has developed an advanced AI algorithm that successfully identifies and separates voices based on their spatial location in the room. This innovative approach uses information about how sounds are reflected within a room, allowing the isolation of desired voices even in complex, chaotic environments. The goal is to filter out everything that is not coming from the speaker’s position, which significantly improves the quality of the audio recording.
This tool has become crucial in forensics, where it is of great importance to identify exactly who is speaking in recordings used as evidence. The first practical application was in a homicide case in the US, where the technology helped turn previously unusable audio recordings into key evidence that led to a conviction.
McElveen’s company also sees huge potential in the military industry, where the technology is used to analyze sonar signals, but also in future commercial applications such as smart speakers, car audio, augmented reality (AR) and virtual reality (VR). Thanks to this technology, devices will be able to better understand human speech in noisy environments, which will improve the user experience in everyday interaction with technology.
In addition to forensics, AI is already helping with other aspects of audio analysis. For example, technology like Bosch SoundSee uses sound signals to predict malfunctions in machines, while AI models in forensic analysis help recognize speakers and verify the authenticity of audio recordings.
Interestingly, during research, it was discovered that the Wave Sciences algorithm uses mathematical principles very similar to those used by the human brain to solve the “cocktail party problem”. This opens up new opportunities for research and improvement of acoustic and cognitive technologies, both for forensics and for everyday use.
This technology is a major step forward in solving one of the most complex problems in acoustics, with wide application in various industries and further potential for the development of advanced sound systems.
The post How AI solves the “Cocktail Party Problem” – The future of forensics and sound technologies appeared first on ITNetwork.
Source: www.itnetwork.rs
