• btc = $68 001.00 2 236.13 (3.40 %)

  • eth = $3 284.74 111.34 (3.51 %)

  • ton = $6.74 0.13 (1.94 %)

  • btc = $68 001.00 2 236.13 (3.40 %)

  • eth = $3 284.74 111.34 (3.51 %)

  • ton = $6.74 0.13 (1.94 %)

25 Sep, 2023
1 min time to read

Scientists have unveiled a speaker equipped with seven "self-deploying" microphones that can partition a room into distinct "speech zones."

This innovation allows the speaker to track and identify different voices even as they move around the room. Moreover, the researchers claim that this precise localization empowers the speaker to not only separate simultaneous conversations but also mute noisy zones or individuals on command, particularly useful for video conferences and meetings.

At the base of this innovative speaker design is a so-called "robotic acoustic swarm". The thimble-sized robot microphones communicate with each other, autonomously moving like miniature Roombas to different points in the room and returning to the charging station when needed. The ability to navigate and rearrange the microphones is essential to achieve accurate voice separation.

The prototype robots use a technique resembling high-frequency echolocation to orient themselves in space. By strategically placing microphones in the room correctly, the neural network processing the data can perform highly accurate calculations. However, the current version of the robots can only localize voices in two-dimensional space and moves across tabletops.

Lead author Malek Itani from the Paul G. Allen School of Computer Science & Engineering described the innovation, stating, "For the first time, using what we're calling a robotic 'acoustic swarm,' we're able to track the positions of multiple people talking in a room and separate their speech."

Experiments in real conditions confirmed the system's capabilities. Testing was conducted in a variety of environments, including offices and kitchens, with three to five participants.

Remarkably, the system, with no prior knowledge of the voice's location or settings, located the voice within 1.6 feet of each other 90% of the time, and the average error was just under six inches.

This page contains "inserts" from other sites. Their scripts may collect your personal data for analytics and their own internal needs. The editorial board recommends using tracker-blocking browsers to view such pages. More →