AI Headphones Isolate Voices in Crowd

Researchers at the University of Washington have developed an innovative AI system that enables noise-canceling headphones to isolate and amplify a single speaker's voice in noisy environments, allowing users to focus on specific individuals by simply looking at them.

Target Speech Hearing System

The Target Speech Hearing (TSH) system, developed by University of Washington researchers, leverages AI to modify the auditory perception of headphone wearers based on their preferences. The proof-of-concept device, presented at the ACM CHI Conference on Human Factors in Computing Systems, is not yet commercially available, but the researchers have made the code publicly accessible on GitHub for others to build upon.

• The TSH system builds upon the team's previous work on "semantic hearing," which allowed users to select specific sound classes to focus on while canceling other sounds in the environment.
• Although the current system can only enroll one speaker at a time and requires the target speaker to be the loudest voice present during enrollment, the researchers are working on expanding the technology to earbuds and hearing aids in the future.

How TSH Technology Works

The TSH system operates by equipping off-the-shelf headphones with microphones and an onboard AI neural network. To isolate a specific speaker's voice, the user simply needs to gaze at the intended speaker for three to five seconds while pressing a button on the headphones. This initiates the "enrollment" process, during which the headphones capture sound signals from the targeted speaker.

• The AI analyzes these signals in real-time to identify the speaker's unique vocal patterns, which are then sent to another neural network tasked with continuously separating the targeted voice from background noise.
• Once enrolled, the system can "latch on" to the speaker's voice, allowing the user to hear the isolated voice clearly even when moving around or facing away from the speaker.

User Testing and Feedback

University of Washington researchers tested the TSH system with 21 subjects, who reported an average improvement in the clarity of the targeted speaker's voice by nearly twice as much compared to unfiltered audio. While the current enrollment process only works if the target speaker is the loudest voice in the room, users can re-enroll the speaker to improve sound quality if they are not satisfied with the initial results. The technology has the potential to revolutionize communication in various settings, such as museums, city streets, and even in brand-name headphones and earbuds.

Future Enhancements and Applications

The TSH system has the potential to significantly benefit individuals with partial hearing loss or those who frequently find themselves in noisy environments where holding conversations can be challenging. Researchers are optimistic about future modifications to address current limitations, such as the ability to enroll multiple speakers simultaneously and isolate voices in more complex auditory settings. Plans are underway to adapt this technology for use in earbuds and hearing aids, which could greatly expand its accessibility and functionality for a wider range of users.

Follow us:

• Website: www.care-investments.com ??
• MSI Partners: www.msi-partners.de ??
• AI Group: Learn more ??
• M&A Group: Learn more ??
• MSI Partners on LinkedIn: MSI Partners ??

Visit our partner page: MSI Partners ??

#AI #AiHeadphones #TechNews #AiNews