Hey HN! I wanted to share this OSS project called Yap — a tiny menu-bar app for macOS 26 that does voice-to-text for any input field. Set a hotkey, speak, press again, and text pastes instantly. Everything runs locally using Apple's new SpeechAnalyzer and SpeechTranscriber APIs; no cloud, no API keys, no account, and the app ships zero models.

A recent benchmark put Apple's on-device model at 2.12% word error rate on clean audio and 4.56% on noisy, beating Whisper Small (3.74% / 7.95%) while running roughly 3x faster across 5,559 LibriSpeech clips. The whole app is about 3,000 lines of native Swift in a 4 MB binary that idles near 60 MB RAM with zero network code.

Features include a global rebindable shortcut (default ⌘⇧D), live waveform and partial transcript, local searchable history, and paste-via-clipboard that works even in Chromium apps. Requires macOS 26 (Tahoe) or later; earlier versions lack the OS-managed models. Permissions needed: Microphone, Speech Recognition, Accessibility, and Automation.

Built by FrigadeHQ, MIT licensed, and fully open at github.com/FrigadeHQ/yap. Install via Homebrew (`brew install --cask frigadehq/tap/yap`) or grab the signed .dmg from Releases. No roadmap bloat — just a single, lightweight tool that stays out of your way.