Every pillar, one tag. The range, not a skills list.
Transcription, diarisation, and speech running entirely on Apple silicon, and why keeping voice local is a product decision before it is a technical one.
A 600M-parameter local model beats Whisper Large v3 on a Mac. The backend choice is per-model, decided by latency and memory and GPU contention, not ideology.
An on-device speech stack for Apple Silicon, in Swift.
whisper_schedule