Seeking to improve physical AI-to-speech applications, sensiBel and Aizip are working to combine sensiBel’s optical microelectromechanical systems (MEMS) microphone technology with Aizip’s edge AI beamforming.
The goal is to improve speech clarity and recognition for more accurate and natural human-AI interaction with cloud AI or large language models.
“The best way to improve AI audio capture at the application level is to use the highest-performance microphone at the input level,” said Kieran Harney, CEO, sensiBel.
Harney said the combination of the low-latency, low-power audio beamforming and noise cancellation solution from Aizip with the MEMS microphones from sensiBel offers a high-fidelity acoustic sensing layer for physical AI.
Aizip’s beamforming technology integrates AI noise reduction, dereverberation and directional speech enhancement to increase speech quality to automatic speech recognition engines, the company said. This works in noisy or far-field environments.
Aizip tested the technology claiming a more than 30% reduction in character error rate for accurate transcription and understanding in physical AI applications. When combined with the optical MEMS microphone, it can identify the speaker’s direction and position as well as the near-field/far-field acoustic scenarios, the companies claim.
