Development of a Low-Cost Voice Announcement System Using Google Text-to-Speech, Raspberry Pi, Python, and SQLite

Abstract

This paper presents a practical and integrated solution for developing an automated voice announcement system using Google Text-to-Speech (TTS), TensorFlow Lite, a Raspberry Pi 5, the Python programming language, and a local SQLite database. The system is designed to provide a lightweight, low-cost platform capable of generating real-time audio announcements, based on stored data or triggered events, suitable for environments such as smart homes, public spaces, or industrial monitoring. The core of the system runs on a Raspberry Pi 5, which serves as both the processing unit and the audio output hub. Python is used to orchestrate all system components, including database access, text retrieval, TTS requests, and audio playback. The system retrieves text messages from a local SQLite database, converts them into audio using the Google Cloud Text-to-Speech API, and plays the resulting speech through a speaker. In addition to cloud-based speech synthesis, the system incorporates TensorFlow Lite to perform lightweight, on-device inference. This enables the system to support local decision-making, pre-processing of input, or even context-aware responses without relying entirely on internet connectivity. For example, TensorFlow Lite can be used to detect patterns in input data, classify message categories, or optimize the timing of announcements. Test results show that the system performs reliably under real-time conditions with minimal latency. Google TTS provides high-quality, multilingual voice synthesis, while TensorFlow Lite enhances the system’s intelligence and responsiveness at the edge.

Received: 05 July 2025 / Accepted: 30 August 2025 / Published: 25 September 2025