Overview
HLK-V20 is the latest high-performance pure offline speech recognition module for a large number of pure offline control scenarios and products. It can be widely and quickly applied to smart homes, various smart small appliances, 86 boxes, toys, lamps, industrial, medical, Products that require voice control, such as the Internet of Things, automobiles, security and lighting. 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. The audio signal is trained and learned through the neural network to improve the voice signal Recognition ability. The solution supports offline recognition of 150 local commands, can freely customize wake-up words, command words and response broadcast words, and has a wealth of peripheral interfaces. Add a relay to the bottom plate to control high-power electrical appliances.
HLK-V20 is the latest high-performance pure offline speech recognition module for a large number of pure offline control scenarios and products. It can be widely and quickly applied to smart homes, various smart small appliances, 86 boxes, toys, lamps, industrial, medical, Products that require voice control, such as the Internet of Things, automobiles, security and lighting. 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. The audio signal is trained and learned through the neural network to improve the voice signal Recognition ability. The solution supports offline recognition of 150 local commands, can freely customize wake-up words, command words and response broadcast words, and has a wealth of peripheral interfaces. Add a relay to the bottom plate to control high-power electrical appliances.
Functional characteristics
Processing kernel:
●32bit RISC core, running frequency 240M
● Support DSP instruction set and FPU floating point arithmetic unit
● FFT accelerator: supports up to 1024 points of complex FFT/IFFT operations or 2048 points of real FFT/IFFT operations Calculate
● Customized speech algorithm operator
Storage:
● Built-in high-speed SRAM
● Built-in 2MB FLASH
Audio input and output:
● Support 1 analog Mic input
● Support dual channel DAC output
● Support I2S input/output
Hardware introduction and description
Board size ：31*75mm
Weight ：36g
Board size ：31*75mm
Weight ：36g
