FluidVoice v1.6.0

What's New

Fills a missing piece in local dictation apps — on-device AI runtime for smart formatting, context-aware capitalization, and post-processing, all running locally with zero data leaving your Mac
Specifically trained on 100,000+ dictation data points for accurate, context-aware enhancement
Requires ~3.5 GB disk space for the model and ~3.5 GB RAM while running
Simpler setup with clear model download progress and warm start for lower first-run latency
Optional report button in history to flag when Fluid Intelligence gets it wrong, so we can keep improving

Parakeet is even faster with near-instant text insertion and zero delay
Dictation formatting split into four toggles: lowercase first letter, remove trailing period, space between dictations, and smart capitalization based on the focused field
Faster recording and processing overlays

Refreshed onboarding flow with language-first voice engine setup, real dictation tryout, and AI enhancement setup in one pass
Apple Speech voice engine option with language-aware setup and no-download fallback
Cleaner permission guidance and tighter low-resource handling

Optional local dictation audio history with budget controls and ZIP export — find it in Preferences → Audio History
Fixed budget pruning for restored history with missing or orphaned audio files

Unified Cohere and Nemotron language selection inside the active speech model row
Made launch window display optional for login-item launches
Some feedback system improvements

Thanks to @postoso for making the launch window optional for login-item launches.
Thanks to @EliseiNicolae for improving the Pill recording overlay visuals.