DFRobot - Gravity: Offline Voice Recognition Sensor (121 Built-in Commands + Custom Commands)

The Offline Speech Recognition Module is an efficient solution ideal for human-computer interaction, machine learning, voice assistants, and smart home voice control projects. It is easy to use and compatible with Arduino, micro:bit, and ESP32, with detailed tutorials provided. It offers plug-and-play functionality through I2C and UART. The module supports recognition of 121 built-in English command words, with additional customizable commands. (e.g., recording a cat's meow for an automatic feeder). This module is designed for command recognition (keyword-based), not full speech-to-text or AI conversation. It is highly integrated, including an onboard speaker and microphone, providing real-time voice feedback.

Easy to Use

The offline voice module is simple and compatible with existing platforms like Arduino UNO, micro:bit, and ESP32, with detailed Arduino IDE tutorials and Makecode tutorials. It supports plug-and-play through I2C and UART communication methods, allowing easy integration into projects.

121 Ready-to-Use Words

The built-in commands are commonly used in educational and smart home scenarios, allowing for immediate use without any configuration, such as "Play music," "Open the door," "Turn on the light," and "Close the window." Here is the full list of command words.

Self-Learning Function

The self-learning function is user-friendly, allowing users to add 17 customized command words. Any language or sound can be trained as a command, providing great flexibility for interactive audio projects.

Teaching the device new commands is simple. Activate the learning mode with the wake-up word, then say "Learning command word," and repeat the desired command three times. After these steps, the device will have learned the new command. Detailed instructions are provided to ensure an easy learning process.

No Internet Connection, More Secure

Users need not worry about their conversations being recorded and uploaded to the cloud. The module functions seamlessly in outdoor projects, schools, and other environments with unstable networks. Unlike existing online voice assistants that rely on cloud platforms and require internet access, potentially compromising device data privacy during use.

High Integration, Saving More Space

The offline voice module integrates an onboard speaker and microphone. High integration saves space and eliminates the hassle of complicated wiring, enhancing the project's neatness and stability.

Real-Time Voice Feedback

Users can immediately receive recognition results during use, improving user experience and convenience.

The Offline Speech Recognition Module is an easy-to-use, highly integrated solution perfect for various projects, offering 121 built-in commands, 17 customized commands, real-time voice feedback, and secure offline operation without the need for an internet connection.

Version History & Important Information

V1.1 (May 8, 2025):

This version includes minor iterative refinements focused on physical handling and assembly convenience:

• Improved Speaker Connection: The onboard speaker connection is an easier plug-and-play connector (V1.0 has a soldered connection). This simplifies assembly and modification.
• Enhanced Sensor Mounting: Mounting holes have been added to the sensor board for more secure attachment.

V1.0 (March 2023):

The original robust and fully capable module. It provides the complete suite of offline speech recognition features and offers a highly effective and reliable user experience.

Version Update of Gravity: Offline Language Learning Voice Recognition Sensor

• User-Friendly: Plug and play, compatible with Arduino UNO, micro:bit, and ESP32.
• 121 Pre-Programmed Commands: Ready for immediate use without any setup.
• Self-Learning Capability: Supports the addition of 17 customized command words.
• Offline Operation, Enhanced Privacy: No network required.
• High Integration: Includes an onboard speaker and microphone.
• Instant Voice Feedback: Provides real-time recognition results.

• Voice recognition interaction
• Voice-controlled terminal
• Competition and project development

• Operating Voltage: 3.3 - 5V
• Maximum Operating Current: ≤370 mA (5V)
• Communication: I2C/UART
• I2C Address: 0x64
• Fixed Commands: 121
• Fixed Wake-up Command: 1
• Custom Commands: 17
• Learning Activation Command: 1
• Onboard Microphone Sensitivity: -28dB
• Module Size: 49×32 mm / 1.93×1.26”
• Operating Temperature: 0-70℃

• Gravity: Voice Recognition Module - I2C & UART (V1.1) x1
• Gravity-4P I2C/UART Sensor Connector x1