南宫28ng

Home > Campany News

Actions Technology Releases Another Edge AI Audio Chip: ATS362X's Low Power Consumption and High Computing Power Ignite a New Wave in Audio

03/10/2025

As Deepseek triggers another wave of innovation, Actions Technology responds to the rapid expansion of artificial intelligence from cloud to edge computing. The ATS362X edge AI chip, an important member of Actions Technology's edge AI audio chip series, is now officially launched. Targeting AI entertainment audio devices, professional audio equipment, and AIoT edge computing terminals, this chip delivers a tri-core heterogeneous architecture, 24-bit lossless audio quality, and an exceptional energy efficiency ratio of 6.4 TOPS/W, providing powerful capabilities for both consumer and professional audio AI upgrades.

配图1-En

The ATS362X, as Actions Technology's new-generation tri-core heterogeneous SoC chip equipped with AI-NPU (Actions Intelligence NPU), adopts an ARM STAR CPU + HIFI5 DSP + MMSCIM NPU architecture design. This chip will inject AI capabilities into various edge audio products, helping terminal brand products enter the new AI era. Currently, the chip solution is in the collaborative development stage with multiple brand clients and is expected to be implemented in terminal products in the near future.

 

Superior AI Computing Power, Enabling Real-Time Local Inference

As a flagship AI audio chip from Actions Technology, the ATS362X employs an advanced, highly elastic tri-core heterogeneous architecture. The HIFI5 and MMSCIM work collaboratively to achieve a theoretical computing power of 132 GOPS, supporting real-time edge inference for complex models such as voiceprint recognition and environmental sound classification. The MMSCIM core, after sparse model optimization, can improve from 6.4 to 19.2 TOPS/W@INT8, with operating power consumption at only milliwatt levels. Compared to traditional processor architectures, it offers significant improvements in practical computing power and energy efficiency. Additionally, the chip is fully compatible with mainstream AI frameworks including TensorFlow, PyTorch, and ONNX, and provides ANDT toolchains and a rich neural network operator library, significantly enhancing model deployment efficiency. The Soft SIMD technology supports custom operator development, flexibly adapting to cutting-edge requirements such as voice interaction and multimodal perception, greatly expanding product application boundaries and providing comprehensive support for brand clients in developing personalized AI applications.

 

Professional Audio Quality, Meeting High-Fidelity Requirements

The ATS362X demonstrates excellent audio performance. The chip integrates 4-channel 24-bit ADCs (SNR>111dBA) and 2-channel 24-bit DACs (SNR>113dBA), with total harmonic distortion plus noise (THD+N) as low as -100dB, delivering audio quality comparable to professional recording equipment. The chip supports multi-channel I2S (16 channels @384KHz) and ASRC dynamic sampling rate adjustment (with performance reaching THD+N less than -140dB), enabling precise detail capture in complex sound fields and fully meeting the high-fidelity requirements of professional stage speakers, digital mixing consoles, and similar applications.

 

Rich Interfaces and Security Design, Ensuring Product Safety and Reliability

In terms of connectivity, the ATS362X integrates high-speed interfaces including SPI (96MHz), USB 2.0 HS, and SD/EMMC, meeting the requirements for multi-device interconnection and expansion. For security design, the chip features built-in hardware-level AES-256/SHA-256 encryption, RSA2048 signature, and TRNG random number generation functions, providing comprehensive protection for data privacy and device security.

配图2-EN

Jeffers Gong, General Manager of the Portable Audio Business Unit at Actions Technology, stated: "The launch of the ATS362X represents an important milestone for Actions Technology in the field of edge AI audio. By integrating high-performance AI computing power, professional-grade audio processing capabilities, and flexible interface design, we are committed to providing customers with smarter, higher-quality, and more personalized AI audio solutions, driving rapid upgrades and iterations across the entire industry to fully embrace the new AI era."