Generative Language Model Technology Integrated into an IoT Device for the Development of a Voice Assistant

The integration of artificial intelligence technologies into IoT devices has opened new possibilities for interaction with the environment through voice assistants, such as ChatGPT, improving interaction with smart devices in sectors such as home, health, and education. However, the adoption of these technologies faces challenges due to device heterogeneity, the need for interoperability, and concerns about data privacy and security. The objective of this research is to develop an IoT device that integrates artificial intelligence technologies and generative language models for a voice assistant, covering the design of a voice recognition system, the implementation of efficient communication with the model, the coordination between ESP32 microcontrollers and the integration of a voice synthesis system. The results show that the system can send queries to ChatGPT and receive responses in real time, validating its ability to handle natural language processing. Furthermore, speech synthesis, using Audio.h library and the MAX98357 module, have demonstrated effective text-to-audio conversion, while the integration of the INMP441 microphone and the Google Cloud Speech-to-Text platform ensures voice capture and processing. In conclusion, the operation of the IoT device and its real-time interaction with the ChatGPT API were validated to obtain an efficient text-to-speech conversion, being scalable for future improvements.