Features: Used for large number of offline control scenarios and product. The audio- signal is trained and learned through the neural-network to improve voice signal Recognition ability. Supports offline recognition of 150 local commands, freely customizable wake-up word, command word and response broadcast word, and has a wealth of peripheral- interfaces. The module uses a 32bit RSIC architecture core, and adds a DSP instruction set specifically for signal processing and speech recognition, an FPU- arithmetic- unit that supports floating-point operations, and an FFT- accelerator. Can be widely and quickly applied to intelligent home, various small appliances, 86 boxes, lamps, industrial, Product that require voice control, such as the Internet of Things, automobiles, security and lighting. Specifications: Material: Metal Size: As Shown Package Includes: 1 Set Controller note: Processing kernel: 32bit RISC core, running frequency 240M Support DSP instruction set and FPU- floating point operation unit FFT- accelerator: supports up to 1024 points of complex FFT-/IFFT- operations or 2048 points of real FFT-/IFFT- operations Storage: ●Support 1 analog- Mic- input ●Support dual-channel DAC output ●Support 12S input/output Peripheral- interface: ●Support up to 10 GPIO ●All GPIO can be configured as external interrupt- input and call source ●1 standard SPI-Master interface, the highest speed is 30MHz- ●1 SPI- Slave interface with a maximum rate of 30MHz- ●1 full-duplex UART maximum rate 3Mbps ●I I2C master-/slave controller with a maximum speed of 400kHz ● 2 PWM outputs ● 1 12-bit SAR-ADC maximum 450Khz sampling rate