The Secret Behind ‘Furiosa’: AI Blends Anya Taylor-Joy and Child Star’s Faces!
May 29, 2024
Inside the $12M AI Project That Will Redefine Computing: University of Texas to Revamp Operating Systems to Do More with Less Energy
May 29, 2024
Your audial experience may become otherworldly…thanks to this new innovation using AI.
AI-Powered Noise-Cancelling: A New Frontier
Modern life is filled with noise, and traditional noise-cancelling headphones reduce all sounds indiscriminately. A new AI system, Target Speech Hearing, aims to change that by allowing users to focus on a specific voice, keeping it audible while cancelling out other noises. Developed by researchers at the University of Washington, this technology is currently a proof of concept but has potential applications in popular noise-cancelling earbuds and hearing aids.
Efficient and Adaptive Technology
The challenge of separating human voices requires complex neural networks, which need to operate in real time with limited computing power and battery life. The researchers addressed this by using knowledge distillation, a technique that compresses large AI models into smaller, efficient ones. This smaller model can extract specific vocal patterns, enabling the Target Speech Hearing system to function effectively in commercial headphones. Activation involves an “enrollment” process, where the user records a sample of the target speaker’s voice, allowing the system to isolate and prioritise that voice even amidst background noise.
Practical Applications and Future Development
Currently, the system can only enroll a speaker whose voice is the loudest, but improvements are underway to enhance its capabilities. Experts like Sefik Emre Eskimez from Microsoft and Samuele Cornell from Carnegie Mellon University recognize the practical potential of this technology, especially in scenarios like meetings where isolating a single voice can be crucial. This innovation represents a significant step towards practical applications of speech separation research, providing a new tool for enhanced communication in noisy environments.
(Visit MIT Technology Review for the full story)
*An AI tool was used to add an extra layer to the editing process for this story.
