Ai-Thinker VC-02-Kit — Offline Voice Recognition Development Board with Mic & Speaker
Offline voice recognition just became accessible to every Indian maker, student, and product developer — the Ai-Thinker VC-02-Kit is a complete, ready-to-use development board built around the US516P6 voice processor from Unisound, delivering pure offline speech recognition with up to 150 custom voice commands, no internet connection required, no cloud dependency, no latency, and no privacy concern. Say a command, get an instant response — entirely on-device.
Whether you are building a voice-controlled home automation relay, a hands-free toy, a smart lamp, or a prototype for an industrial IoT product, the VC-02-Kit gives you everything in one box: the development board, a microphone, and a speaker. Plug in a USB cable, upload your firmware or use the factory commands, and your project is talking back to you in minutes.
What Makes It Stand Out
Most voice recognition solutions for makers either require a cloud connection (introducing latency, privacy risks, and internet dependency) or involve complex ML model training on powerful hardware. The VC-02-Kit takes a different approach entirely — it runs a deeply optimised, chip-integrated offline recognition algorithm on the US516P6’s 32-bit RISC core running at 240 MHz, with a dedicated DSP instruction set and FFT accelerator (supporting up to 1024-point complex FFT/IFFT operations) specifically tuned for speech processing. The recognition happens in under 100ms, locally, with no external compute.
The onboard CH340C USB-to-serial chip means you can connect directly to a PC for firmware flashing, serial debugging, and UART communication — no separate USB-to-TTL adapter needed. Three onboard LEDs (React in blue for wake state, Warm in orange, Cool in white) give you instant visual feedback aligned to the factory firmware’s voice commands, making it extremely satisfying to demo and debug. Two buttons — RST for reset and IOA26 for manual wake-up — round out the developer experience.
What truly sets this board apart is Ai-Thinker’s custom firmware platform at voice.ai-thinker.com — a browser-based tool where you can define your own wake words, command phrases (in Chinese or English), and output actions, then generate and flash a custom firmware binary without writing a single line of C code. For Indian makers building commercial products, this is a game-changer: you can prototype and iterate voice command sets in hours rather than days.
Why Makers Choose This Offline Voice Recognition Board
The UART interface (UART1 at 115200 bps default) means you can connect the VC-02-Kit to any microcontroller — Arduino, ESP32, STM32, or Raspberry Pi — and receive hex command codes over serial whenever a voice command is recognised. No complex library integration needed; your microcontroller just reads the serial output and maps hex codes to actions.
The factory firmware ships with 150 pre-programmed voice commands covering common smart home scenarios — lights, fans, air conditioners, and more — available in the factory command list PDF. For a complete hands-on getting started walkthrough with wiring, firmware, and first command demo, watch the official VC-02-Kit tutorial video which covers everything from unboxing to first voice response.
The module also supports I2C for connecting sensors and displays, PWM for motor and LED control, and DAC for audio output routing — giving you a surprisingly rich peripheral ecosystem for a pure voice recognition module.
Technical Specifications
Here are the fully verified datasheet specifications for this offline voice recognition development board:
| Parameter | Value |
|---|---|
| Board Model | VC-02-Kit |
| Module | VC-02 |
| Main Chip | US516P6 (Unisound) |
| CPU | 32-bit RISC core @ 240 MHz |
| DSP | Dedicated DSP instruction set for speech/signal processing |
| FFT Accelerator | Up to 1024-point complex / 2048-point real FFT/IFFT |
| SRAM | 242 KB high-speed built-in |
| Flash | 2 MB built-in |
| Voice Commands | Up to 150 offline commands |
| Language Support | Chinese and English |
| Microphone Input | 1× analog mic, SNR ≥ 94 dB |
| Audio Output | Dual-channel DAC |
| Power Supply | 5V, current > 500 mA |
| Active Current | 56.1 mA (min) – 232 mA (max) |
| Standby Current | ~56 mA |
| Hibernation Current | 0.6 – 0.7 mA |
| UART1 Default Baud | 115200 bps (max 3 Mbps) |
| UART0 Default Baud | 57600 bps |
| Interfaces | UART, I2C (3.3V & 5V), PWM, DAC, GPIO |
| GPIO | 5 external GPIOs |
| Package | DIP-19, 2.54 mm pin header |
| Board Size | 42.2 × 35.6 mm (±0.2 mm) |
| USB Interface | CH340C USB-to-serial (onboard) |
| LEDs | React (blue), Warm (orange), Cool (white) |
| Buttons | RST (reset), IOA26 (manual wake-up) |
| Operating Temperature | –40°C to +85°C |
| Storage Conditions | –40°C to +125°C, <90% RH |
Full electrical characteristics and pinout are available in the VC-02-Kit specification PDF.
Best Used For
The VC-02-Kit is purpose-built for five categories of projects where offline voice recognition adds immediate, tangible value.
Smart home voice control is the most natural fit. Connect the VC-02-Kit’s UART output to an ESP32 or Arduino that controls relay modules, and you have a fully offline voice-controlled switch panel for lights, fans, and appliances — no Wi-Fi, no Alexa, no subscription. The factory firmware already includes commands for common appliances, making this a weekend build for most makers.
Voice-controlled toys and educational products are an ideal application for the 150-command capacity and low cost. Build a talking robot that responds to “move forward,” “stop,” and “turn left” purely offline, with instant response and zero cloud lag. The kit’s compact form factor and complete mic+speaker inclusion make it easy to embed into enclosures.
86-box panels and wall-mounted smart switches — a market segment Ai-Thinker explicitly designed the VC-02 for — benefit from the module’s small footprint, 5V operation, and UART output that integrates cleanly with panel MCUs. Indian makers building custom smart home panels or retrofitting existing switches can use this board as a drop-in voice input layer.
IoT voice trigger nodes that send MQTT messages or HTTP requests via an ESP32 companion can use the VC-02-Kit as the voice front-end — the ESP32 listens on UART, receives the command hex code, and takes network action. This keeps the voice processing entirely local even in an internet-connected system, protecting user privacy and ensuring the device still works during an outage.
Custom firmware product prototyping using the Ai-Thinker voice platform lets engineers and product companies build proof-of-concept voice UI for appliances, industrial controls, and consumer devices without embedded ML expertise. Define your wake word, your commands, your output mappings — and get a flashable binary in minutes.
Frequently Asked Questions
Q1. Does this offline voice recognition module need an internet connection to work?
No — that is the core advantage of the VC-02-Kit. All speech recognition processing runs entirely on the US516P6 chip onboard. There is no Wi-Fi, no cloud API call, no internet dependency of any kind. The module works in remote locations, areas with poor connectivity, and scenarios where user privacy is a priority. Offline voice recognition means the device responds even when your internet is down.
Q2. How do I connect the VC-02-Kit to an Arduino or ESP32?
Connect VCC (5V) and GND from your power supply to the kit, then connect TX1 of the VC-02-Kit to the RX pin of your Arduino/ESP32, and RX1 to the TX pin. Configure your serial port at 115200 bps. When the module recognises a voice command, it sends a unique hex code over UART that your microcontroller reads and maps to an action. No special library is needed — standard serial reading works perfectly.
Q3. Can I create my own custom voice commands?
Yes. Ai-Thinker provides a free browser-based custom firmware tool at voice.ai-thinker.com where you can define your own wake words and command phrases in Chinese or English, assign output actions, and generate a firmware binary. Flash this binary to the VC-02-Kit via USB (using the CH340C interface) and your custom offline voice recognition set is live immediately.
Q4. What is included in the VC-02-Kit package?
The kit includes the VC-02-Kit development board, one microphone module, and one speaker — everything you need to start testing offline voice recognition without sourcing additional audio hardware. You only need a USB cable and a PC to get started.
Q5. What languages does this offline voice recognition module support?
The US516P6 chip and VC-02 firmware support both Chinese and English voice commands. For Indian makers, English command sets work well for standard control phrases like “turn on the light,” “turn off the fan,” and similar appliance commands. Custom firmware allows you to define the exact English phrases your project needs, with the recognition engine optimised to understand them accurately.
What’s in the Box
- 1 × Ai-Thinker VC-02-Kit Development Board
- 1 × Microphone Module
- 1 × Speaker
Related Products
- Ai-Thinker VC-01-Kit Offline Voice Recognition Board
- ESP32 DevKit V1 Development Board
- 5V Single Channel Relay Module
Important Links
🔗 VC-02-Kit Specification PDF — Full datasheet with pinout, electrical characteristics and dimensions
🔗 Factory Voice Command List — Complete list of 150 pre-programmed voice commands in factory firmware
🔗 Custom Firmware Builder — Official Ai-Thinker browser tool to create your own wake words and commands
🔗 Getting Started Video — Step-by-step setup tutorial for the VC-02-Kit
Verified owner Sachin Pathave (verified owner) –
Product received in appropriate packaging. Fast shipping and delivery. Your YouTube channel really helpful and inspiring.
Store manager techiesms –
Glad to hear that you loved our product and service
Happy Making 😇
Verified owner Ishan Kesarwani (verified owner) –
Best for voice control and is very easy to use by seeing the youtube videos made by them.
Verified owner Chirrenthen (verified owner) –
Very nice product!!. The packaging was good without any damage. The videos about this module is very helpful and easy to learn 😀. Thankyou so much @techiesms !
Store manager techiesms (verified owner) –
Glad to hear the you liked the product
Happy Making!!
Shravani7 –
very good product!! nice customer service as well~
should have added the data sheet of the product in the package. but over all nice experience