Swift Scribe - AI-Powered Speech-to-Text Private Transcription App for iOS 26 & macOS 26+

Real-time voice transcription, advanced speaker diarization, on-device AI processing, and intelligent note-taking exclusively for iOS 26 & macOS 26 and above

Uses Apple's new Foundation Model Framework and SpeechTranscriber. Requires macOS 26 to run and compile the project. The goal is to demonstrate how easy it is now to build local, AI-first apps.

⚠️ This is mostly to demonstrate the abilities of the new APIs, we don't plan to actively maintain the project. Over time we may integrate new features as a way to demonstrate abilities from FluidAudio. ⚠️

🎯 Overview

Swift Scribe is a privacy-first, AI-enhanced transcription application built exclusively for iOS 26/macOS 26+ that transforms spoken words into organized, searchable notes with professional-grade speaker identification. Using Apple's latest SpeechAnalyzer and SpeechTranscriber frameworks (available only in iOS 26/macOS 26+) combined with FluidAudio's advanced speaker diarization and on-device Foundation Models, it delivers real-time speech recognition, intelligent speaker attribution, content analysis, and advanced text editing capabilities.

🛠 Technical Requirements & Specifications

System Requirements

• iOS 26 Beta or newer (REQUIRED - will not work on iOS 25 or earlier)
• macOS 26 Beta or newer (REQUIRED - will not work on macOS 25 or earlier)
• Xcode Beta with latest Swift 6.2+ toolchain
• Swift 6.2+ programming language
• Apple Developer Account with beta access to iOS 26/macOS 26
• Microphone permissions for speech input

🚀 Installation & Setup Guide

Development Installation

1. Clone the repository:

   git clone https://github.com/seamlesscompute/swift-scribe
   cd swift-scribe

2. Open in Xcode Beta:

   open SwiftScribe.xcodeproj

3. Configure deployment targets for iOS 26 Beta/macOS 26 Beta or newer

4. Build and run using Xcode Beta with Swift 6.2+ toolchain

⚠️ Note: Ensure your device is running iOS 26+ or macOS 26+ before installation.

📋 Use Cases & Applications

Transform your workflow with AI-powered transcription:

Business & Professional

• 📊 Meeting transcription with automatic speaker identification and minute generation
• 📝 Interview recording with real-time speaker diarization and attribution
• 💼 Business documentation with speaker-tagged content and report creation
• 🎯 Sales call analysis with participant tracking and follow-up automation

Healthcare & Medical

• 🏥 Medical dictation and clinical documentation
• 👨‍⚕️ Patient interview transcription with medical terminology
• 📋 Healthcare report generation and chart notes
• 🔬 Research interview analysis and coding

Education & Academic

• 🎓 Lecture transcription with chapter segmentation
• 📚 Study note creation from audio recordings
• 🔍 Research interview analysis with theme identification
• 📖 Language learning with pronunciation feedback

Legal & Compliance

• ⚖️ Court proceeding transcription with timestamp accuracy
• 📑 Deposition recording and legal documentation
• 🏛️ Legal research and case note compilation
• 📋 Compliance documentation and audit trails

Content Creation & Media

• 🎙️ Podcast transcription with automatic speaker labeling and show note generation
• 🎬 Video content scripting with professional speaker diarization
• ✍️ Article writing from multi-speaker voice recordings
• 📺 Content creation workflows with speaker-attributed production notes

Accessibility & Inclusion

• 🦻 Real-time captions for hearing-impaired users
• 🗣️ Speech accessibility tools with customizable formatting
• 🌐 Multi-language accessibility support
• 🎯 Assistive technology integration

🏗 Project Architecture & Code Structure

Scribe/                     # Core application logic and modules
├── Audio/                  # Audio capture, processing, and FluidAudio speaker diarization
├── Transcription/         # SpeechAnalyzer and SpeechTranscriber implementation
├── AI/                    # Foundation Models integration and AI processing
├── Views/                 # SwiftUI interface with rich text editing
├── Models/                # Data models for memos, transcription, speakers, and AI
├── Storage/               # Local data persistence and model management
└── Extensions/            # Swift extensions and utilities

⭐ Advanced Features

Speaker Diarization

• FluidAudio Integration: Industry-grade speaker identification and clustering
• Research-Grade Performance: Competitive with academic benchmarks (17.7% DER on AMI dataset)
• Real-time Processing: Live speaker identification during recording with minimal latency
• Speaker Attribution: Color-coded transcription with confidence scores and timeline mapping

Speaker Identification

• Automatic Speaker Detection: No manual configuration required
• Speaker Persistence: Consistent speaker identification across recording sessions
• Visual Attribution: Rich text formatting with speaker-specific colors and metadata
• Speaker Analytics: Detailed insights into speaking patterns and participation

Privacy-First

• Fully On-Device: All processing happens locally - no cloud dependencies
• Zero Data Transmission: Audio and speaker data never leave your device
• Secure Storage: Speaker embeddings and models stored securely with SwiftData
• Complete Offline Operation: Works without internet connectivity

🗺 Development Roadmap & Future Features

📄 License & Legal

This project is licensed under the MIT License - see the LICENSE file for complete details.

🙏 Acknowledgments & Credits

• Apple WWDC 2025 sessions on SpeechAnalyzer, Foundation Models, and Rich Text editing
• Apple Developer Frameworks - SpeechAnalyzer, Foundation Models, Rich Text Editor
• FluidAudio - Professional speaker diarization and voice identification technology