Fluenti: AI Powered Speech Therapy & Emotional Support

Abstract

The barriers to speech therapy and emotional support also include increased accessibility and inconsistent quality, relating to both children with speech disorders and adults who need mental health assistance. Such issues can be tackled through this project the creation of Fluenti, which is the union of interactive story building games that can be used in speech therapy with AI oriented emotional support sessions aimed at adults. The architecture uses a multi layered design where the frontend will be based on React and TypeScript and the backend will be based on Node.js and Express.js, with the specialized therapist services being based on Python Flask. In the case of speech therapy, Google Gemini creates age specific interactive narratives and carries out evaluations in four types of therapies, which are pronunciation, fluency, Developmental Language Disorder (DLD), and social communication. The story game uses Web Speech APIs browser native and speech recognition in real-time, dynamically changing difficulty level, and progress tracking, which is gamified. To offer emotional support, a Python-based therapy bot, which is driven by Groq LLM, offers responses with hybrid crisis detection (AI and pattern-based) and psychological profiling and session management. The site combines OpenAI Whisper API with speech-to-text and OpenAI TTS with text-to-speech to allow the voice responses. MongoDB takes care of user progress, session history, and psychological profiles data persistence. The security features are JWT authentication, two-factor authentication, encryption and rate limiting. The system is a connection between therapeutic interventions and modern technology, which offers accessible, scaled, and individual support. The further development will focus on a broader range of types of therapies, improved crisis-detecting algorithms, and the system modification to multilingual support and global markets.