🎙️ AudioTranscriber – iOS App with Whisper Transcription

An iOS app built using SwiftUI and AVFoundation to record audio in real time, segment it into 30-second chunks, and transcribe the segments using OpenAI Whisper API. The app supports background recording, audio monitoring, offline queuing, and a full SwiftData-based session management system.

---

🚀 Features

• Real time audio recording using AVAudioEngine
• Automatic 30-second segmentation
• Whisper API integration for transcription
• Retry logic with exponential backoff for failures
• Offline queuing and retry
• Background recording support
• Audio quality selection (low, medium, high)
• Real-time input volume visualization
• Data persistence using SwiftData (RecordingSession, Segment, Transcript)
• VoiceOver accessibility for visually impaired users
• Pause/Resume recording, audio quality selector, and secure HTTPS requests
• Local fallback using Apple’s SFSpeechRecognizer if Whisper fails
• Fully tested UI with pagination, pull-to-refresh, and interruption handling
• Cloud-safe secure API key handling

---

🧱 Architecture

• AudioRecorder.swift – Core audio engine logic
• ContentView.swift – Main UI for controls, volume meter, and navigation
• SessionListView.swift – Past recordings list with pagination
• SessionDetailView.swift – Shows transcript for each segment
• TranscriptionService.swift – Handles communication with OpenAI Whisper
• SwiftData Models (RecordingSession, Segment, Transcript)

---

🛠️ Setup Instructions

1. Clone the Repo

https://github.com/19ska/AudioTranscriber.git

2. Open in Xcode

• Open AudioTranscriber.xcodeproj or AudioTranscriber.xcworkspace in Xcode 15+

3. Add OpenAI API Key

• Create a file named Secrets.plist in the root directory with:

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
<dict>
<key>OpenAIKey</key>
<string>sk-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx</string>
</dict>
</plist>

• Add Secrets.plist to .gitignore

4. Enable Capabilities

• Go to Signing & Capabilities tab in Xcode: ✅ Background Modes → Audio, AirPlay, and Picture in Picture ✅ App Sandbox → Audio Input ✅ Microphone usage description in Info.plist:

   <key>NSMicrophoneUsageDescription</key>
   <string>This app requires microphone access to record audio</string>
  

5. Run on Real Device

• VoiceOver, microphone, and background audio features must be tested on a physical device.
• Build and run via USB or WiFi connection.

---

🧱 Data Model

@Model
class RecordingSession {
    @Attribute(.unique) var id: UUID
    var startTime: Date
    var segments: [Segment]
}

@Model
class Segment {
    @Attribute(.unique) var id: UUID
    var filePath: String
    var timestamp: Date
    var transcript: Transcript?
    var session: RecordingSession
}

@Model
class Transcript {
    @Attribute(.unique) var id: UUID
    var text: String
    var segment: Segment
}

---

Accessibility

• Every interactive control is labeled with accessibilityLabel and accessibilityHint
• Live volume visualization described with percentage
• Session and transcript views are fully accessible

---

💡 Development Notes

• AVAudioSession interruptions and route changes are handled.
• Audio segments are saved every 30 seconds and uploaded independently.
• If network is offline, segments are stored and retried later.
• Uses @Published properties to live update UI.