• Employs BNPU V4.0, supporting neural networks such as CNN, TDNN, GRU, Transformer, and LSTM, as well as parallel vector operations; 

• 32-bit RISC-V CPU, clock speed 73.728 MHz, 256 KB SRAM;

 • Four power consumption modes: Shutdown mode power consumption <2 μA; Normal mode power consumption ranges from 1 to 6 mA;

 • Provides a wide range of peripheral interfaces, including UART, I2C, I²S PWM, GPIO, RTC, SPI, and KEYSCAN; 

 • Application: Low-power, high-performance edge-side voice interaction.

• Employs BNPU V4.0, supporting neural networks such as CNN, TDNN, GRU, Transformer, and LSTM, as well as parallel vector operations; 

• 32-bit RISC-V CPU, clock speed 73.728 MHz, 256 KB SRAM; 

• Four power consumption modes: Shutdown mode power consumption <2 μA; Normal mode power consumption ranges from 1 to 6 mA;

• Provides a wide range of peripheral interfaces, including UART, I2C, I²S PWM, GPIO, RTC, SPI, and KEYSCAN; 

• Applications: Low-power, high-performance deep noise-reduction voice interaction; low-power wake-up and call noise reduction.

• Brain Neural Processing Unit (BNPU) V3.5 

• 32-bit high-performance RISC-V CPU, up to 210MHz, with 288KB SRAM 

• Integrated PMU, RC oscillator, and Audio Codec supporting one Audio ADC and one Audio DAC 

• Supports 1xUART/2xUART/3xUART, 1xI2C, 4xPWM, and multiple GPIO interfaces 

• Supports on-device speech recognition, local command word self-learning, and multi-intent recognition 

• CI13082V supports 3.3V power supply (others support 5V); CI13162P has built-in amplifier (others require external amplifier)