How eatekha/finditapp works
Project type and purpose
• This is an iOS mobile client application (FindIT) built with Swift and SwiftUI.
• It records and processes audio, performs speech recognition (hot-word or continuous), sends/transcribes audio (likely via Whisper or a remote service over WebSocket), and displays relevant “movie” or “item” search results in a SwiftUI interface.
File structure
• Top-level Xcode project/xcworkspace files (FindIT.xcodeproj, FindIT.xcworkspace).
• App source in FindIT/:
• SwiftUI entrypoint: FindITApp.swift, ContentView.swift
• Views: MovieView.swift, MovieDetailView.metal, TestView.swift, ViewController.swift
• ViewModel: MovieViewModel.swift
• Managers: AudioManager.swift, AudioRecorder.swift, SpeechRecognizer.swift, WhisperManager.swift, WebSocketManager.swift
• Assets & Info.plist
• Third-party frameworks in Pods/:
• Cobra-iOS (Picovoice for voice activity detection)
• ios-voice-processor (audio buffering and processing)
• Podfiles, test targets (FindITTests, FindITUITests)
README
• Very minimal (“MyFinance”)—project name placeholder. No embedded architecture details, so infer from code.
Main components and interactions
a. Components
1) UI Layer (SwiftUI Views & ViewController)
2) MVVM Layer (MovieViewModel)
3) Audio Pipeline
• AudioRecorder / AudioManager
• VoiceProcessor (from ios-voice-processor)
• Cobra (hot-word/voice activity detection)
4) Speech-to-Text
• SpeechRecognizer (local or remote STT)
• WhisperManager (handles Whisper API or model)
5) Networking
• WebSocketManager (real-time communication with backend or transcription service)
6) Assets & Rendering
• MovieDetailView.metal (custom Metal shader)
7) External frameworks
• Cobra-iOS xcframework
• ios-voice-processor Swift package
b. Relationships & data flow
1) User taps UI → ContentView calls MovieViewModel
2) MovieViewModel invokes AudioManager/AudioRecorder
3) Audio is buffered → VoiceProcessor → Cobra for VAD/hotword
4) On detection, audio frames go to SpeechRecognizer and/or WhisperManager
5) WhisperManager may use WebSocketManager to stream audio to a remote Whisper service
6) Transcribed text returns to MovieViewModel
7) MovieViewModel queries local or remote movie-search API → updates published properties
8) SwiftUI Views observe ViewModel → render MovieView, MovieDetailView with Metal shader
c. Architectural patterns & principles
• MVVM: clear separation between Views and business logic
• Modular audio pipeline: pluggable components (record → process → recognize)
• Dependency injection/facade for external services (WebSocketManager, WhisperManager)
• Use of Swift packages and CocoaPods for decoupled third-party code
d. Key technologies & frameworks
• SwiftUI (UI)
• Combine or ObservableObject (data binding)
• CocoaPods (Cobra-iOS, ios-voice-processor)
• WebSocket for real-time service communication
• Metal shading (MovieDetailView.metal)