Changelog

What's new in Speakmac. Follow along as we ship improvements.

Latest

Version 3.0.0 – Dictation reliability overhaul

  • Stop/start dictation is much more reliable in live preview mode, with session-bound cleanup to prevent stuck overlays and missed final text.
  • Final pasted text now always comes from full batch transcription after you stop recording, improving consistency over streaming edge cases.
  • Model download onboarding is clearer: progress now switches to a proper preparing state after download, and avoids misleading stalled percentages.
  • Fresh installs no longer trigger unnecessary Documents permission prompts.

Version 2.9.3 – More reliable history & storage fixes

  • Your Speakmac history now migrates safely to Application Support on upgrade, avoiding “missing history” issues.
  • If your local transcription database gets corrupted, Speakmac can automatically recover (and backs up the old files).
  • Smoother transcribing indicator animation with less background churn.

Version 2.9.2 – Dictation polish & smarter paste hints

  • Dictation now handles newline punctuation more naturally, keeping commas/periods in the right place.
  • Paste hints are smarter about focus and custom editors, reducing false warnings.
  • New dictation command corpus test helps prevent formatting regressions.

Version 2.9.1 – More reliable install attribution

  • Install attribution now recovers the download identifier from your Downloads folder for more reliable campaign tracking.
  • Removed an experimental WhisperKit integration that wasn’t production ready.
  • Cleaner release packaging with fewer extra install artifacts.

Version 2.9 – Dictation commands & stability fixes

  • Dictation commands let you say "comma", "period", "new line", "question mark" and 20+ other formatting commands while speaking—just like Apple Dictation but for Speakmac.
  • Fixed a crash that could occur on fresh installs during update checks.
  • Fixed speech model downloads occasionally failing silently—now completes reliably.

Version 2.8 – One-click activation & regex replacements

  • One-click license activation—click the link in your purchase email and Speakmac activates instantly.
  • Regex word replacements now available for power users—match patterns like "U S A" → "USA" with regular expressions.
  • Better geo insights help us understand where users need regional pricing (with your consent).

Version 2.7 – Analytics improvements

  • Better insights into how you use Speakmac—hands-free mode, live preview, and auto-submit usage are now tracked (with your consent) to help us improve.
  • Update tracking now properly identifies which version you upgraded to, helping us ensure smooth rollouts.

Version 2.6 – Hands-free mode & universal paste

  • Lock mode for hands-free dictation—press Space during recording to keep it running without holding your hotkey.
  • Auto-paste now works in Bike, Obsidian, Notion, VS Code, and 50+ more apps that use custom text editors.
  • Redesigned language packs UI with clearer download status and hotkey help.
  • Privacy mode banner reminds you when analytics are off—dismiss it anytime.
  • Coaching hints guide you through lock mode with gentle prompts.

Version 2.5 – Live preview & word replacement

  • Live preview shows your words appearing in real-time as you speak—watch transcription happen before your eyes.
  • Custom word replacement fixes names that get misrecognized—add "Cindy → Sydney" and it's corrected automatically.
  • Press Shift during recording to toggle auto-submit on or off—your preference persists across sessions.
  • Simplified engine: FluidAudio is now the only offline provider. No more confusing Apple Speech vs FluidAudio choice.
  • Smarter voice detection waits for you to start speaking before transcribing, reducing false starts.
  • FluidAudio upgraded to 0.8.0 for improved offline accuracy.
  • Fixed auto-submit sending wrong key in Terminal apps.

Version 2.4 – Shift auto-submit & dock toggle

  • Hold Shift while using push-to-talk and Speakmac will press Enter for you after pasting—perfect for quick sends in Slack or Messages.
  • New visibility picker in Settings lets you choose Dock + Menu Bar, Menu Bar Only, or Dock Only—whichever fits your workflow.
  • Streamlined settings with a new Developer sidebar for power users.

Version 2.3 – Chinese language support

  • Chinese transcription is here. Download the language pack in Settings → Language and switch between Multilingual (25 languages) and Chinese with one click.
  • Supports Mandarin, Cantonese, Sichuan, Henan, and Tianjin dialects with high accuracy.
  • Updates now install silently in the background—your app stays current without interrupting your flow.
  • Cleaner sidebar with icon-only navigation and instant tooltips.

Version 2.2 – Menu bar polish & conversion insights

  • Your hotkey now appears in the menu bar next to "Start Recording", just like native macOS shortcuts.
  • Settings reorganized into Help & Guides, Software Update, and Contact Support sections.
  • Privacy mode indicator updated to a lock icon with "Private" label for clarity.
  • Trial includes 30 free transcriptions (previously 40).
  • Fixed text insertion not working when Speakmac was the frontmost app.

Version 2.1 – Faster downloads & smoother setup

  • Speech model downloads are now 4-6x faster.
  • Mini player mode—a compact floating waveform for distraction-free dictation.

Version 2.0 – Hotkey improvements & updates

  • Left and right modifier keys (⌘, ⌥, ⌃, ⇧) now register as separate bindings so you can set distinct hotkeys for each hand.
  • Silent background updates now work reliably—no more stale versions sitting in /Applications.
  • Added a Help Center with step-by-step guides for setting up hotkeys and customizing your workflow.

November 2025 – Onboarding & privacy tune-up

  • Setup now includes a live practice field so you can try dictation before leaving onboarding.
  • Analytics stay off until you flip the consent toggle that appears after your first demo.
  • There is a single analytics switch inside Settings → Privacy whenever you want to change your mind.

October 2025 – Reliable streaming & smarter audio

  • Low-level HAL capture keeps USB, Bluetooth, and studio mics connected without dropouts.
  • Streaming preps Fluid Audio, Apple Speech, or Deepgram ahead of time so sessions start instantly.
  • Sub-second or silent takes auto-delete so your history only shows real recordings.

October 2025 – Offline Parakeet release

  • Onboarding now spells out Apple Speech vs Parakeet so you can pick the offline path you want.
  • Parakeet downloads once (about 600 MB), caches locally, and is always warmed up for instant use.
  • Fluid Audio is the default engine on fresh installs so private, offline dictation works out of the box.

More updates coming soon. Follow @kirjd for the latest.

Changelog

Get the latest version

Download Speakmac and get every update automatically.

Requires macOS 14+ and Apple Silicon.