Description
The VC-02 Voice Control Module is an offline speech recognition sensor designed to control electronic devices using predefined voice commands. It does not require internet or AI processing, making it ideal for embedded systems, automation projects, and low-power applications.
It can recognize a limited set of trained voice commands and trigger digital outputs or communicate with a microcontroller (such as Arduino) via UART.
The VC-02 is an offline speech-recognition module developed by Shenzhen Ai-Thinker, built around the US516P6 voice chip from Unisound. It’s tailored for low-cost, standalone voice control in smart devices like lamps, toys, appliances, and smart home systems.
| Parameter | Details |
|---|---|
| Manufacturer | Ai-Thinker |
| Core Chip | US516P6 (Unisound) |
| Processor | 32-bit RISC core @ 240 MHz, DSP instruction set, FPU, FFT accelerator |
| Memory | 242 KB SRAM, 2 MB Flash |
| Voice Capability | Up to 150 offline commands (Chinese & English) |
| Audio Input | 1 × analog mic (SNR ≥ 94 dB), up to 4 × digital mics |
| Audio Output | Dual DAC outputs, I²S interface |
| Interfaces | UART (up to 3 Mbps), SPI, I²C, ADC, PWM, GPIO |
| Power Supply | 3.6 V – 5 V |
| Clock Sources | 12 MHz RC oscillator, PLL system clock |
| On-Chip Features | LDO for 3.3 V & 1.2 V, POR, low-voltage detection, watchdog timer |
| Form Factor | Module: small SMD; Kit: 42.2 × 35.6 mm (USB + indicators) |
| Development Kit | USB-to-Serial (CH340C), status LEDs, easy prototyping |
| Applications | Smart home devices, robotics, appliances, toys, IoT voice control |
| Limitations | Custom firmware tool needed; risk of bricking if flashing incorrectly |
“VC-02: A compact offline voice recognition module capable of processing up to 150 commands without cloud connectivity — ideal for smart devices, robotics, and IoT projects.”
Features
* Support bilingual control, both Chinese and English
* Single MIC Access
* Support AEC echo elimination, steady-state noise reduction
* Support to wake up from learning, no need to compile firmware
* Comprehensive recognition rate can reach more than 98%
* Identification time is less than 100ms
* Extremely low error rate
* Entry corpus up to 150Kernel Introduction
* Integrated 32bit RISC kernel, frequency up to 240MHz
* Support for DSP instruction sets and FPU floating point arithmetic units
* FFT Accelerator: Maximum Support 1024 points FFT / IFFT operation, or 2048 points FFT / IFFT operation
* Unisound Equation Customized Logo Algorithm
* Built-in 242KB high speed SRAM
* 8kb ROM for boot
* Built-in 2MB SPI Flash
* Support 1 road simulation MIC input, SNR ≥ 94dB
* Support for dual channel DAC output
* Built-in 5V to 3.3V, 3.3V 1.2V LDO is power supply for chip
* Provide complete RTOS-based SDK
* Support bilingual control, both Chinese and English
* Single MIC Access
* Support AEC echo elimination, steady-state noise reduction
* Support to wake up from learning, no need to compile firmware
* Comprehensive recognition rate can reach more than 98%
* Identification time is less than 100ms
* Extremely low error rate
* Entry corpus up to 150Kernel Introduction
* Integrated 32bit RISC kernel, frequency up to 240MHz
* Support for DSP instruction sets and FPU floating point arithmetic units
* FFT Accelerator: Maximum Support 1024 points FFT / IFFT operation, or 2048 points FFT / IFFT operation
* Unisound Equation Customized Logo Algorithm
* Built-in 242KB high speed SRAM
* 8kb ROM for boot
* Built-in 2MB SPI Flash
* Support 1 road simulation MIC input, SNR ≥ 94dB
* Support for dual channel DAC output
* Built-in 5V to 3.3V, 3.3V 1.2V LDO is power supply for chip
* Provide complete RTOS-based SDK
Demo Video :
https://www.youtube.com/watch?v=MKJ2mlHhuNk
No products in the cart. 
Reviews
There are no reviews yet.