NXP releases SAF9xxx, upgrading automotive AI audio processing!

NXP Semiconductors has released the SAF9xxx series, which brings multiple artificial intelligence audio functions to the in-vehicle infotainment system, which is an important technical improvement in automotive audio processing. The newly launched audio digital signal processing ( DSP ) solution is designed to meet the growing demand for artificial intelligence audio functions in software-defined vehicles ( SDVs ). SAF9xxx uses Cadence's new generation of high-performance Tensilica HiFi 5 DSP and combines it with a dedicated neural network engine to efficiently realize the next generation of high-quality self-learning audio and voice applications.

In addition, the SAF9xxx series can integrate up to 5 tuners through software definition, covering global mainstream wireless broadcast applications such as DAB, HD-Radio, DRM, CDR and AM/FM. Both audio and wireless broadcast applications of car entertainment systems can be completed with this single chip and are easy to expand.

Significance

Audio performance is a major differentiator for in-vehicle infotainment systems, and it also provides OEMs with new opportunities to differentiate mid-range and high-end vehicles. The integration of SAF9xxx’s machine learning capabilities with neural networks opens up new areas of functionality, such as identifying and distinguishing drivers and passengers by tone and accent. Machine learning algorithms can detect emergency vehicle sirens, analyze the vehicle’s location and dynamics, and provide relevant audio-visual information to the driver.

“NXP’s SAF9xxx DSPs are off-the-shelf infotainment solutions designed with AI/machine learning processing to deliver richer functionality and features, as well as better performance to meet the demanding audio and wireless requirements of today’s and tomorrow’s software-defined cars,” said John van den Braak, senior vice president and general manager of audio systems at NXP. “The scalable architecture makes it cost-effective and easy for OEMs to deploy.”

More details

The SAF9xxx series is ideal for low-latency audio algorithm applications, such as real-time active cancellation of engine noise, road noise, wind noise, and tire noise.

Built-in fast memory access and comprehensive analog and digital I/O further enhance the performance of the Gigahertz-class DSP to support increasingly demanding audio functions.

Gigabit Ethernet enables fast connection directly to the car network.

NXP's extensive audio IP ecosystem partners have a wealth of audio software and algorithms and have planned to develop new audio applications based on SAF9xxx.

Availability and Support

The SAF9xxx series will debut at the Audio Engineering Society's fifth annual International Conference on Car Audio , June 26-28, 2024 .