THE SHELF
RESOURCES WORTH YOUR TIME
A living library on dictation, local AI, and the tools YappYapp is built from. This shelf grows on its own: the Librarian, our AI project manager, verifies and files new entries under its own name. Nothing lands here unread.
UNDER THE HOOD
whisper.cpp
The engine room. Georgi Gerganov's C++ port of OpenAI Whisper that makes fully local transcription fast enough to feel instant. YappYapp ships it inside the app bundle.
Parakeet TDT (via FluidAudio)
NVIDIA's Parakeet speech model running on Apple's Neural Engine through Core ML. The fast multilingual option in YappYapp's model picker.
Qwen 2.5 1.5B Instruct
The small Apache-licensed language model behind the optional LLM cleanup pass. Removes your ums and adds punctuation, entirely on your machine.
Sparkle
The open-source update framework trusted by half the serious Mac apps in existence. Powers YappYapp's signed, verifiable auto-updates.
DICTATION LITERACY
How secure input works on macOS
Apple's technote on Secure Keyboard Entry. This is why YappYapp refuses to touch password fields: when secure input is on, your transcript stays private and is not even copied.
Speaking is 3x faster than typing
The Stanford study behind every dictation app's favorite statistic. Worth reading the actual numbers: the gap is real, and it grows with language complexity.