Anker Unveils AI Chip for Smarter Audio Devices

Anker has unveiled its proprietary Thus processor, a custom silicon chip engineered to embed local artificial intelligence directly into audio devices, mobile accessories, and IoT products. This innovative chip represents a significant advancement as the world's first neural-net compute-in-memory AI audio chip. It boasts a smaller form factor and reduced power requirements compared to conventional chips, making it ideal for integration into smaller devices.

The core innovation of the Thus chip lies in its unique architecture, which places computation directly where the AI model resides. This eliminates the need for constant data transfer between storage and processing units, a bottleneck in existing AI chips. Anker CEO Steven Yang explained that this design prevents the AI model from needing to move, enhancing efficiency for every inference.

Soundcore's forthcoming flagship earbuds will be the debut device for the Thus chip. Anker selected earbuds due to their extreme size constraints and limited power availability. Previous designs could only accommodate small neural networks with limited parameters, but Thus's energy-efficient compute-in-memory design supports millions of parameters, enabling sophisticated AI functions like advanced noise cancellation.

This enhanced processing power, combined with eight MEMS microphones and two bone conduction sensors, is expected to deliver significantly cleaner call audio in the new Soundcore earbuds. Anker aims to overcome the limitations of traditional noise-canceling technology, which struggles in noisy environments. The company will officially present full product details and additional AI-powered features at Anker Day on May 21.