Show HN: NitNab a macOS 26 Privacy Centric AI Transcription App (github.com)

NitNab is an open-source, privacy-first macOS transcription app built for macOS 26+ and Apple Silicon that combines Apple’s on-device Speech framework and Apple Intelligence (FoundationModels) to transcribe, summarize and interact with audio locally. It supports common audio formats (M4A, WAV, MP3, FLAC, etc.), batch processing, real-time progress, auto-saving to a structured folder or iCloud, and exports to TXT, SRT, WebVTT, JSON and Markdown. The app provides an interactive three-tab UI (Transcript/Summary/Chat) so you can generate AI summaries, ask context-aware questions, draft emails, extract action items, and maintain conversation history — all without sending audio or text to third-party clouds. For the AI/ML community, NitNab shows a practical, production-ready pattern for on-device speech + local LLM-style features using Swift 6, SwiftUI, actor-based services, SFSpeechRecognizer, LanguageModelSession and FoundationModels. It emphasizes privacy, offline capability, and performance on Apple Silicon while demonstrating how modern Swift concurrency (async/await, actors) and MVVM architecture can power safe, extensible AI apps. The project is MIT-licensed and on GitHub, making it a useful reference or starting point for developers wanting to build or research local, privacy-preserving speech-to-text and summarization workflows.