Sparse AI MCU Enhances Edge Voice Processing and Cleanup

Femtosense and ABOV Semiconductor have collaborated to offer a sparse AI MCU, namely AI-ADAM-100, which combines Femtosense's Sparse Processing Unit 001 (SPU-001) with ABOV Semiconductor's MCU. This innovative MCU leverages sparse AI technology, enabling on-device AI functionalities, particularly voice-based control in various appliances and products. By cleaning up voice and audio data before transmission to the cloud, it enhances reliability, accuracy, and reduces the amount of data sent. A key advantage of the AI-ADAM-100 is its ability to implement voice language interfaces at the edge, even without cloud connectivity.

According to Sam Fok, CEO for Femtosense, the integration of sparsity throughout the AI development stack makes AI-ADAM-100 the first device on the market to fully utilize the benefits of sparse AI. Home appliances and small form factor devices like high-fidelity hearing aids, industrial headsets, and consumer earbuds are among the target applications for this sparse AI MCU.

But how does sparse AI function? Through the elimination of irrelevant portions of the algorithm, sparse AI significantly reduces the cost of AI inferencing. Only the nonzero, relevant segments of the algorithm require hardware resources, resulting in a 10x improvement in speed, efficiency, and memory footprint.

AI-ADAM-100 offers advanced wake-up and control features for home appliance manufacturers. By allowing other system controllers and connectivity modules to enter low-power sleep mode when not in use, significant power savings can be achieved. ABOV has also verified the voice command recognition performance of AI-ADAM-100 under varying noise conditions.

In the realm of AI and MCUs, AI-ADAM-100 delivers unique language processing and voice cleanup capabilities coupled with compatibility for manufacturers' own AI models, regardless of their density (dense or sparse). This offering presents exciting prospects for expanding the integration of AI into diverse devices.

Earlier, SSP reported that chip stocks in Asia experienced a significant decline.