v3.23.0

Added

• Settings → Permissions shows live Microphone and Accessibility status, with the same repair steps as first-run when a switch was turned off

v3.22.0

Fixed

• Opening Settings → Models after an update no longer freezes on Checking local model integrity

v3.21.0

Fixed

• First-run no longer marks setup complete when models are already on disk if Microphone or Accessibility is still off
• The download HUD says Models downloaded, not Ready, and does not appear during first-run

Changed

• Launch at login is on by default for new installs
• Settings, Help, and first-run copy is shorter. Accessibility is required for Listen because insert into the active field is on by default

v3.20.0

Fixed

• Long transcribing HUD copy on macOS 13.3 keeps rounded capsule ends instead of clipping to a rectangle

v3.19.0

Fixed

• First-run model download shows progress instead of staying on a spinner
• HUD capsules keep rounded ends on macOS 13.3, including transcribing and unknown-language toasts
• Reading HUD Cancel works on macOS 13.3
• Transcription finishes even if Accessibility is not granted; the system prompt no longer blocks the copy HUD
• History loads on macOS 13.3 instead of staying on Loading History

v3.18.0

Fixed

• Reading HUD no longer crashes when its message updates
• Cancelling a paste restores the previous clipboard contents
• Interrupted model downloads resume from the correct byte offset

v3.17.0

Changed

• Status HUD is closer to the old compact size
• The capsule grows with the message instead of clipping it
• Recording Cancel sits farther from the waveform

v3.16.0

Added

• Status HUD while recording, transcribing, or reading, with Cancel (and pause/resume while speaking)
• Settings → General HUD position (Top or Bottom). New installs default to Top

Changed

• Visual alerts use a glass capsule with a thin colored outline
• Headset pause toggles pause/resume instead of stopping playback
• Leaving the Voices settings tab no longer stops Read playback
• Voices use given names in Settings and on the reading HUD

v3.15.0

Changed

• Settings → Maintenance splits factory reset (erase data, keep the app) from uninstall (erase leftover models/settings, then show the app in Finder so you can move it to the Trash). OneSentence cannot delete itself.
• Quit after factory reset, uninstall, or first-run setup actually quits instead of flashing the button.
• The installer disk image shows drag-to-Applications layout.

v3.14.0

Added

• Insert transcription into the active field after Listen (Settings → General, on by default)
• Recording overlay with input meters when Visual Alerts are on

Changed

• Default speech-to-text engine is Parakeet; Whisper is optional. Switching to Whisper does not start a download until you tap Download on a model
• Whisper acceleration is Metal or Neural Engine (macOS 14+), chosen in Models. Neural uses the Core ML encoder for every listen
• Voice quality is Fast / Balanced / Quality (2 / 8 / 10 flow-matching steps). The old 3–30 slider sat past where this voice improves
• Voice playback primes the audio graph so the first syllable is not clipped
• Whisper language default stays Auto-Detect (or English if auto is missing). Opening Models no longer snaps the language to Bulgarian
• First-run download is about 1 GB (TTS plus the selected STT engine)
• Sparkle waits until first-run setup finishes, and no longer shows a start-up error when XPC helpers live only in the Sparkle framework