Echo 3.0.0

Dictate into any app.

• The Echo keyboard: tap Dictate, Echo starts listening, go back to your app and talk, tap when you are done, and the text lands at your cursor. Transcription still happens on your phone.
• iOS keeps keyboards off the microphone, so the keyboard opens Echo to record in the background. Space, delete, return and the globe key work without Full Access; dictating needs it.
• Not yet tried on a real iPhone. The next build needs the regenerated signing profiles (already made).

Echo 2.5.0

• Meetings on the Mac now run on Apple's speech model too, every line labelled You or Them.
• The app is 143 MB smaller: Whisper base no longer ships. Small is more accurate (1.9% of words wrong against base's 2.2%) and still ships with tiny.
• Before iOS 26, iPhones with 4 to 6 GB of memory now get tiny in Auto (6.0%) instead of base. Base is still a download under Model in Settings.
• The benchmark now scores the speech, not the reading's spoken credit.

Echo 2.4.0

• Mac: Settings, Meeting, Share and More in one glass bar at the bottom. History is the sidebar, collapsed by default.
• iPhone: a native bottom bar with History, Share, More and Settings, every icon the same size and evenly spaced.
• The record button sits in the page, just above the bar.
• Your model and language choices now stick between launches.
• Benchmarks on a real human voice: Apple 1.5% of words wrong, Whisper small 1.9%, base 2.2%, tiny 6.0%. See docs/BENCHMARKS.md.

Echo 2.3.0

• What's New never clips: standard sheet sizes and a scroll view.
• Speak-Back Voice shows one row per voice, best quality, with the region when a name has two accents.
• Models read Auto, Apple, Tiny, Base, Small, Large.
• The account's sign-in token is this-device-only.
• Apple first: engine-level tests, including a stop pressed while Apple's session is still starting. A failed start no longer leaves a listener hanging.
• Benchmarks: Apple 65x real time, Whisper tiny 57x, base 51x, small 16x on a Mac mini M4 (docs/BENCHMARKS.md).
• TranscriptionEngine split by area into Models, Live, Apple and Files.

Echo 2.2.0

• Apple first: on iOS 26 and macOS 26, live speech runs through Apple's on-device SpeechAnalyzer when it knows your language. Echo asks the system to install Apple's model in the background. Whisper stays as the fallback for older systems, Mac meetings, custom words and other languages.
• Optional account: sign up, confirm, sign in, forgot password, profile name, sign out and delete account, all in Settings. Echo still works fully without one, and transcripts never leave the device.
• New privacy policy and a reset-password page at echo.jaybulb.com.
• Review fixes: bigger Settings tap target, solid glass record button, the live word's spring now shows.
• New tests: real speech through Apple's live session, account error messages.

Echo 2.1.0

Echo is the name again (formerly Notate).

• Jony Ive pass: native nav bar, one record button front and centre, no splash, no cards, no mode switch. Files come in through More, drag and drop on the Mac, or Open With.
• Liquid Glass on iOS 26 and macOS 26.
• The word being heard pops in as you talk.
• History with native search, Settings as one Form.
• Mac: dropping two files with the same name no longer transcribes the old one.
• New icon, brand kit and landing at https://echo.jaybulb.com.
• Docs in docs/, license now Apache 2.0.

The App Store listing still says Notate until Apple takes the new name on the next editable version.

voxprint v1.3.7

Full Changelog: https://github.com/nulljosh/voxprint/commits/v1.3.7