🤯 AI Audio Revolution: Soundcore's Thus Chip 🚀

THE INNOVATION: ANKER’S “Thus” CHIP
Anker has unveiled a groundbreaking custom silicon processor dubbed “Thus,” representing a significant leap in localized AI processing for audio devices, mobile accessories, and IoT applications. This neural-netcompute-in-memoryAI audio chip distinguishes itself through its remarkably compact design and reduced power consumption, making it ideally suited for smaller, resource-constrained devices. CEO Steven Yang highlighted the fundamental shift in chip architecture, stating that existing AI chips traditionally require constant data transfer between model storage and computation, leading to inefficiencies. The “Thus” chip, conversely, integrates the computation directly within the memory where the model resides, eliminating this overhead and dramatically improving performance.

COMPUTE-IN-MEMORY: A REVOLUTIONARY APPROACH
The core innovation of the “Thus” chip lies in its compute-in-memory design. This approach fundamentally changes how AI processing is handled. Instead of moving vast amounts of data back and forth between the chip's memory and processing units – a process that consumes significant power and introduces latency – the “Thus” chip performs computations directly within the memory itself. This eliminates the need for constant data transfers, resulting in a substantial reduction in power consumption and a significant increase in processing speed. The ability to handle several million parameters, a substantial increase over previous designs limited to a few hundred thousand, allows for more complex AI tasks, such as advanced world noise cancellation. This represents a crucial advancement in audio processing technology, directly addressing limitations found in traditional approaches.

EARBUDS AS THE INITIAL TESTBED
Anker has strategically chosen earbuds as the initial platform for integrating the “Thus” chip, recognizing the sector's particularly stringent constraints regarding size and power. The limited space within earbuds poses significant challenges, restricting the available power and necessitating reliance on smaller, less powerful neural networks. Previous designs were constrained by the need for onboard neural networks capable of handling only a few hundred thousand parameters. However, the “Thus” chip's enhanced energy efficiency enables it to manage several million parameters, unlocking substantially greater computing power. This expanded capability is particularly beneficial for features like advanced call noise cancellation, allowing for significantly cleaner audio quality in diverse and challenging environments. The combination of the “Thus” chip, eight MEMS (micro-electromechanical systems) microphones, and two bone conduction sensors promises to dramatically improve voice isolation and audio clarity, regardless of ambient noise levels.