I built Dictly, a macOS app that turns your voice into styled text in near real-time (~200ms latency). Unlike most dictation tools, everything runs 100% on-device: no cloud, no servers, no internet required.

Main features: • Instant streaming transcription — words appear as you speak • Privacy-first — all processing happens locally on Apple Silicon • ⌨ Quick Capture anywhere — press a hotkey and dictate into any text field, system-wide • Custom dictionaries — teach Dictly your jargon, names, or code • Customizable post-processing pipelines — remove fillers, auto-format lists, adjust tone, or apply regex rules in real time

It’s available now on TestFlight with the official release pending on Apple's approval!.

I’d love feedback — especially from developers, writers, and productivity nerds. I’m particularly curious how people feel about the pipeline customization (per-word, sentence, or full transcript).

Link: https://dictly.app

Thanks!

Best regards, Jannik