GonzalezMada/Voice-Controlled-Light-with-Malagasy-Commands-TinyML-

# 🎤Voice-Controlled Light with Malagasy Commands & TinyML **Control lights using voice commands in Malagasy**, a TinyML project deployed on Arduino Nano 33 BLE Sense. *"Mirehitra" (Light On) | "Maty" (Light Off)* --- ## 🌟 Project Overview This project demonstrates how to train a **Convolutional Neural Network (CNN)** to recognize Malagasy voice commands and deploy it on a microcontroller using **TensorFlow Lite Micro**. The AI model processes audio signals to control a light bulb via Malagasy keywords, bridging language gaps in voice technology. **Key Features**: - 🗣️ Supports **Malagasy language** (under-resourced in voice tech). - 🔋 Runs on a **coin-cell battery** with Arduino Nano 33 BLE Sense. --- ## 🛠️ Hardware & Software **Hardware**: - Arduino Nano 33 BLE Sense Rev2 (with built-in microphone). - Light bulb - Relay module - 5V-6V battery **Software**: - Google Colab (GPU for training) - TensorFlow - Arduino IDE --- ## 📂 Project Structure ```bash ├── Training/ │ └── train_voice_control_model_with_malagasy_words.ipynb # Model training & quantization ├── voice_command_based_on_TinyML/ └── voice_command_based_on_TinyML.ino # Arduino deployment code └── arduino_audio_provider.cpp └── arduino_command_responder.cpp └── arduino_main.cpp └── audio_provider.h └── command_responder.h └── feature_provider.cpp └── feature_provider.h └── main_functions.h └── micro_features_micro_features_generator.cpp └── micro_features_micro_features_generator.h └── micro_features_micro_model_settings.cpp └── micro_features_micro_model_settings.h └── micro_features_model.cpp └── micro_features_model.h └── recognize_commands.cpp └── recognize_commands.h