Features
Neural network processor for the audio(A-NPUTM): A-NPUTM supports neural networks such as DNN, TDNN, RNN, and CNN, as well as parallel vector operations. Support speech recognition, Voiceprint Recognition, edge-side NLP, command word self-learning, voice detection;
CPU and memory: The CPU clock speed can reach 180 MHz; Built-in 4MBytes Flash memory; Built-in 480KBytes SRAM; Built-in 512-bit eFuse, can be used for application encryption;
Audio Codec: High-performance, low-power audio ADC, SNR ≥ 95dB; Low-power audio DAC, SNR ≥ 95dB;
Audio Interface: 1-way IIS interface, supports configurable master-slave; 1-channel dual-path PDM interface;
ADC and PWM: Built-in 4-channel 12-bit SAR ADC; Supports 6 PWM interfaces;
GPIO: 26 high-speed GPIOs with a response rate of up to 20MHz; Among them, 18 GPIOs support 5V input;
Reset and power management: Built-in Power Management Unit (PMU); PMU input voltage range: 3.6V to 5.5V; Built-in Power-On Reset (POR); Built-in Voltage Detection (PVD);
Clock: Built-in RC oscillator, also supports external crystal oscillator; developers can choose to use either the built-in RC or an external crystal as the chip clock source depending on different application scenarios;
Communication interface: 1-channel IIC interface; 3 UART interfaces, supporting 5V and up to 3Mbps speed;
Timers and Watchdogs: Built-in 4 sets of 32-bit timers and 2 sets of watchdogs;
WBQFN 5mmX5mmX0.85mm-40L package
Description
AWA89501 is Awinic's high-performance intelligent voice chip, integrating the self-developed A-NPU^TM and CPU core, with a main frequency of 180MHz. It features 480KB of built-in SRAM and rich peripherals such as PMU and audio codecs, requiring only minimal external components to build a high-cost-performance intelligent voice solution.
The chip meets industrial-grade standards, with an operating temperature range of -40°C to 85°C, and complies with multiple reliability certifications such as MSL3, 2KV contact discharge, FCC EMC, ROHS, and REACH.
It’s A-NPU^TM technology supports mainstream neural networks such as DNN, RNN, and CNN, features voice/voiceprint recognition, edge-side NLP, command word self-learning, voice detection.
