Echo: A Voice and Text-Driven Framework for Automated Mobile Application Development Using Large Language Models

The rise of Large Language Models (LLMs) has significantly transformed software engineering by automating and accelerating the processes of application design, development, and deployment. This study presents Echo, a conversational framework that converts voice- and text-based user requirements into production-ready mobile applications. Unlike traditional no- code platforms, Echo generates industry-standard React Native source code, enabling seamless cross-platform deployment on both Android and iOS from a unified code base. The system inte- grates multimodal input processing, iterative intent clarification, structured specification generation, and autonomous build or- chestration. Experimental evaluation shows that Echo drastically reduces the overall application development lifecycle compared to conventional workflows while achieving high intent resolution accuracy even under noisy acoustic conditions. Moreover, the framework preserves the complete developer control over the generated source code, ensuring flexibility, transparency, and extensibility. Index Terms—Large language models, AI-driven program- ming, mobile application generation, natural language interfaces, automated deployment, voice-controlled development, agentic systems