VoiceScribe
Offline voice notes app with on-device transcription and AI-style summaries: overview, key takeaways and action items extracted on device with Apple's NaturalLanguage framework. Folders, favorites, PDF export, review reminders. SwiftUI + SwiftData, zero dependencies.
VoiceScribe is a complete native iOS app source template for recording voice notes, transcribing them on device and summarizing them with Apple's NaturalLanguage framework - no API keys, no backend, no third-party dependencies. Record with a live waveform, get an editable summary with overview, key takeaways and action items, organize notes into folders, export to PDF and ship it under your own brand.
Highlights
- Record voice notes with a live waveform recording screen and duration timer
- On-device speech recognition via Apple's Speech framework (US English by default, one locale identifier to change)
- Extractive AI-style summaries: sentence scoring picks the Overview, Key Takeaways and parsed Action Items - all editable afterwards
- 4-page onboarding flow with a privacy pitch and a permissions dashboard, including recovery alerts that deep-link to iOS Settings
- Folder organization with icon and color pickers, favorites and full-text search
- Audio player tab with waveform visualizer, seeking and playback speed control
- PDF export of transcripts and summaries via UIGraphicsPDFRenderer
- Local notification review reminders scheduled from the note share menu
- Designed empty state plus demo seed data for a great first-run experience
- Orphaned audio file sweep at startup keeps storage clean
- 100% offline: recording, transcription setup, summarization, PDF export and notifications all use frameworks that ship with iOS
- One-file design system: change the accent gradient in Theme.swift and the whole app re-skins itself
Technology
- SwiftUI + SwiftData, iOS 17.0+, iPhone portrait
- Speech framework (SFSpeechRecognizer) with on-device recognition where available
- NaturalLanguage framework extractive summarizer behind a swappable SummaryGenerating protocol
- AVAudioRecorder / AVAudioPlayer with @Observable-era view models
- Zero third-party dependencies: no CocoaPods, no SPM packages, no sign-in
- Pre-configured Info.plist: microphone and speech recognition usage strings, audio background mode
Requirements
- macOS 14.0+ and Xcode 15.0+ (Swift 5.9+)
- iOS 17.0+ device or simulator; a physical iPhone is required for recording and transcription testing
- Apple Developer account only for device testing and App Store submission
What You Get
- Complete Xcode project source: models, views, services, view models and unit tests
- Swappable summary engine: conform one type to SummaryGenerating to plug in a remote LLM backend without touching any view code
- documentation.html: setup, bundle ID, rebranding, re-skin, extension points, troubleshooting and App Store submission checklist
- README.md with a quick-start and project overview
- 11 app screenshots
- Full rights to rebrand, reskin, extend and publish on the App Store
Please note: this is a source code template, not a compiled app. A Mac with Xcode 15+ is required to build it. If you have questions, use the contact form on the site, I usually reply within 24 hours.


