How nisrinelachguer/ai-voicetherapy works
Project type and purpose
• This is a full-stack, three-tier application for automated voice rehabilitation.
• Presentation Layer: Android mobile app (Java, Material Design, Retrofit)
• Business Logic Layer: Spring Boot REST API (controllers, services, JWT security)
• Data Layer: MySQL (JPA/Hibernate) + file storage for audio
File structure
• Frontend (Android)
• app/src/main/java/com/lachguer/ai_voice/ui → Activities & Fragments (UI controllers)
• app/src/main/java/com/lachguer/ai_voice/api → Retrofit ApiService & ApiClient
• app/src/main/java/com/lachguer/ai_voice/model → DTOs and domain models (User, VoiceAnalysis, TherapyExercise)
• app/src/main/java/com/lachguer/ai_voice/utils → AudioConverter, SessionManager (helper classes)
• Resources (layouts, drawables, themes) and Gradle build scripts
• ai_voice (Spring Boot)
• controller → AuthController, VoiceAnalysisController, TherapyExerciseController, UserController
• service → AuthService, VoiceAnalysisService, OpenAIService, TherapyExerciseService, OpenAIConfig
• security → JwtAuthenticationFilter, SecurityConfig, JwtService, UserDetailsServiceImpl
• repository → JPA repositories for User, VoiceAnalysis, TherapyExercise
• dto → Request/Response classes paralleling frontend models
• model → JPA entities
• application.properties and Maven POM
• uploads/audio → persisted audio files awaiting or post-analysis
README for additional insights
• Confirms three-tier architecture and tech stack (Android SDK, Spring Boot, MySQL, JWT, OpenAI Whisper)
• Describes user flow: Registration → Recording → Analysis via Whisper API → Results → Therapy exercises
• Highlights AI integration (speech-to-text, pattern analysis, severity scoring, exercise generation)