Skyworks unveils AI noise suppression solution

Skyworks Solutions, Inc. will demonstrate its AI Noise Suppression solution for removing unwanted background noise from speech signals at CES 2024. Targeting communications and recording systems, the new solution is reported to set a new performance benchmark for noise reduction, increasing speech intelligibility and reducing listener fatigue when using handsets, headsets, conference systems, automotive telematics, wireless microphones and audio-visual systems.

(Source: Shutterstock)

AI noise suppression is commonly used in desktop and cloud applications, but performance lags on the edge, Skyworks said. The new software is designed to be highly scalable from the desktop CPU/GPU down to the edge platform. It is trained on a variety of speech and noise signals and can be implemented on a range of DSP cores and audio frameworks.

“Skyworks is actively developing audio and other signal processing applications at the edge that will replace traditional signal processing techniques and drive new IoT edge applications,” said Yusuf Jamal, senior vice president and general manager of the Diversified Analog Solutions at Skyworks, in a statement. “The AI Noise Suppression solution delivers benchmark performance, scalable from the edge to the cloud while requiring less memory and computational resources.”

The advanced AI is based on a feed-forward neural network that is scalable from the enterprise to the edge. It requires less than 250 kilobytes of memory and adds no latency when integrated into an existing audio processing path, the company said. It is reported to achieve subjective and objective test scores in embedded implementations that compete with desktop CPU/GPU solutions while requiring less than 10% CPU load on a HiFi4 core running at 600 MHz.

Other features include advanced noise reduction on full bandwidth signals in audio/video production and playback, no connection required to large cloud servers or tethering to desktop PCs or mobile phones and no interference with the original speech signal (confirmed by subjective testing based on ITU-T P.808).

The AI Noise Suppression solution is a CES 2024 Innovation Award Honoree and will be demonstrated for the first time at CES 2024 at the Las Vegas Conference Center, North Hall, Booth 9627.