← Back to home

What's new.

Every update to Push to Transcribe, documented.

Need a specific version?

You bought this app, so you get to choose which build you run. If an update ever gives you trouble, install the previous one and keep working.

Loading available downloads. If they do not appear, email [email protected] and we will send you the right installer directly. We hide downloads rather than risk offering a build that has since been withdrawn.
macOS - current
Version 2.0.5 (Build 162)
macOS - previous
Version 2.0.5 (Build 157)

Installing an older build does not stop automatic updates. Push to Transcribe will offer the newest version again unless you turn auto-update off in Settings. Older builds than the ones listed here are not offered: some predate important fixes, and we would rather not hand you a version that causes a new problem while you are solving another one.

Version 2.0.5 (Build 162)
  • License now has its own place in the sidebar, so activating a key takes three steps: open the app, click License, paste your key
  • Dictating in quick bursts no longer drops the start of a recording
  • The activity chart always shows your last seven days
Version 2.0.5 (Build 157)
  • The version number in the sidebar is now a working button: click it to check for updates, or to install one that is already downloaded
  • When macOS’s audio service stops accepting new recordings - which looks like the app breaking for no reason - the app now says so and tells you the one command that fixes it
Version 2.0.5 (Build 155)
  • Polish now gives itself a time budget that grows with the length of your dictation, and turns itself off (and tells you) if your Mac cannot keep up
  • New “Always polish, no time limit” option in Settings > Polish, for long-form dictation
  • Polish only ever uses the engine you selected - if Apple Intelligence cannot polish, you get your own words back right away, never a server you did not choose
  • Llama 3.2 3B is now the recommended LM Studio model
Version 2.0.4 (Build 151)
  • Numbers you say now arrive as digits - “four thousand two hundred dollars” lands as $4,200
  • Your license details are visible in the app, and you can copy your full key with one click to set up another Mac
  • Check for Updates now answers right in the menu instead of opening a separate window
  • Fixed an update check that could stop working after the app had been running a long time
  • New, in beta: Polish - optionally tidy up your transcript with an on-device model. Off by default.
Version 2.0.4 (Build 149)
  • Push-to-talk is dramatically faster - the default recording key now responds in a couple of milliseconds instead of up to a second
  • Recording starts the moment you press it
  • Fixed a menu bar icon that could go blank after your Mac wakes or you change displays
Version 2.0.4 (Build 147)
  • Escape now reliably cancels a recording or a transcription in progress
  • The recorder closes properly instead of appearing stuck
  • No longer switches you to Apple Speech after a single slow moment, usually the first time a model loads
Version 2.0.4 (Build 145)
  • Watches how long transcription actually takes
  • Switches to Apple’s built-in speech engine automatically if your Mac is struggling, and tells you it did
  • Progress counter while the model loads
  • Fixed Check for Updates in the menu bar
Version 2.0.4 (Build 144)
  • Snappier push-to-talk hotkey
  • Fixed intermittent transcription failures
  • No more sluggishness after your Mac wakes from sleep
  • Dictated parentheses are now preserved
  • Live progress counter while transcribing
Version 2.0.3 (Build 136)
  • Refreshed app icon - the new Push to Transcribe microphone mark
Version 2.0.2 (Build 135)
  • Fixes a clipboard issue with Universal Clipboard / Handoff: dictated text no longer overwrites or lingers on your clipboard when syncing between your Apple devices
Version 2.0.1 (Build 134)
  • Fixes a licensing issue that could wrongly show "trial expired" - your license now re-validates and recovers automatically
  • Fixes a rare crash when certain symbols were used in custom word replacements
  • Added in-app help for finding your license key
Version 2.0.0 (Build 133)
  • Bullet lists and "new line" now keep their line breaks instead of collapsing onto one line
  • Lots more emoji by voice - "check", "100", "x", plus 🤣 😭 🥰 🥺 🥳 and more (just say the name + "emoji")
  • Say "insert my phone / email / address" to drop in your saved info, without it firing during normal speech
  • Snappier, more reliable recording and licensing under the hood
Version 2.0.0 (Build 132)
  • New "insert period / comma / colon / dash" voice commands for dictating punctuation
  • Refreshed onboarding with a quick demo video
  • More reliable license checks behind the scenes
Version 2.0.0 (Build 131)
  • The version badge now lights up when an update is ready - click it to update anytime
  • Cleaner filler-word removal: no leftover commas when "um"/"uh" are stripped
Version 2.0.0 (Build 130)
  • New: emoji voice commands - say "fire emoji" for 🔥, "heart emoji" for ❤️, and more
  • ~40 emojis supported, just say the name followed by "emoji"
Version 2.0.0 (Build 129)
  • Much faster paste - transcribed text now lands instantly, no delay after you stop talking
  • Smooth recorder waveform from the first word
  • Version badge is now click-to-copy
Version 2.0.0 (Build 128)
  • Fixed a rare slowdown on the Sound & Mic settings screen after sleep/wake
  • Input level is now checked via System Settings for rock-solid stability
Version 2.0.0 (Build 127)
  • Faster: transcribed text now pastes noticeably sooner after you stop talking
  • Tighter recording-stop sound with less latency
Version 2.0.0 (Build 126)
  • Voice commands no longer corrupt normal speech (colon, period, dash, etc.)
  • Dictation commands now reliable even when you pause mid-phrase
  • One-word answers like "Yes" and "No" no longer dropped
  • Periods and commas land correctly inside quotation marks
Version 2.0.0 (Build 124)
  • Version 2.0.0 - major update with 13 builds of new features
  • Performance: fixed audio meter lag on Sound & Mic settings
  • All sound themes normalized to consistent volume
Version 1.0.0 (Build 121)
  • 5 new sound themes: Kalimba, Kalimba Tone, Marimba, 808 Attack, 808 Tone
  • Real instrument samples - no more generic system sounds
  • Fewer phantom words from trailing silence
Version 1.0.0 (Build 120)
  • 5 new industry dictionary packs: Therapy, Government, Accounting, HR, Insurance
  • Expanded Legal, Medical, and Finance packs
  • 13 packs total with ~2,800 professional terms
Version 1.0.0 (Build 119)
  • Smart Insert now works correctly in Claude Desktop and Electron apps
  • New sentences after periods always capitalize properly
  • Fewer phantom words from accidental hotkey taps
Version 1.0.0 (Build 118)
  • Voice Commands: say "new line", "bullet point", "em dash" and more for instant formatting
  • 19 voice-activated formatting commands
Version 1.0.0 (Build 117)
  • Smart Insert: sentence continuation now keeps trailing punctuation
  • Smart Insert works regardless of Auto-Learn Corrections setting
Version 1.0.0 (Build 116)
  • Smart model selection: new installs get the best model for their hardware
Version 1.0.0 (Build 115)
  • Auto-numbered lists: say "number one... number two..." for formatted lists
  • Line breaks preserved when pasting into Gmail, Notion, Slack
  • Fix: ordinal words no longer falsely trigger list formatting
Version 1.0.0 (Build 114)
  • Auto-numbered lists: say "number one... number two..." for formatted lists
  • Line breaks now work in Gmail, Notion, and other web apps
Version 1.0.0 (Build 113)
  • Dictionary packs expanded to 1,334 entries across 8 industries
  • Legal pack: 236 terms covering Latin, procedures, agencies, constitutional law
  • Medical pack: 230 terms covering drugs, labs, procedures, specialties
Version 1.0.0 (Build 112)
  • Polished recorder pill: app icon with border + red dot side by side
  • Wider pill prevents waveform clipping
Version 1.0.0 (Build 111)
  • Per-app profiles: customize transcription settings for each app you use
  • App icon on recorder pill shows which app you're dictating into
  • Menu bar: click anywhere on last transcription to copy
  • Dictionary: Download Template replaces truncated info panel
Version 1.0.0 (Build 110)
  • Model preloads on app launch - no more waiting on first recording
  • Silence hallucination filter blocks phantom transcriptions from ambient noise
  • Smart Insert now handles mid-sentence replacement after the cursor too
Version 1.0.0 (Build 109)
  • Caps Lock ALL CAPS: turn on caps lock before recording for ALL CAPS output
  • Smart Insert now works in Gmail and Chrome web apps
Version 1.0.0 (Build 108)
  • Smart Insert now works in Gmail, ChatGPT, and all Chrome web apps
  • Mid-sentence insertions auto-strip capitalization and trailing punctuation
Version 1.0.0 (Build 107)
  • Smart Insert: mid-sentence insertions auto-strip capitalization and trailing punctuation
  • Voice Shortcuts onboarding step with email/phone/address setup
  • "Warming up" indicator on recorder pill during model load
Version 1.0.0 (Build 106)
  • Correction toast: larger pill-shaped Yes/No buttons
  • Escape cancel now works during transcription too
Version 1.0.0 (Build 105)
  • Escape key cancels recording without transcribing
  • Shortcuts tab: say a keyword to expand into full text
  • Formatting presets: Default, Email, Notes, Code, Casual
  • No more auto-navigate to Dashboard on recording start
Version 1.0.0 (Build 104)
  • New hotkey: Right Option + Right Shift combo (hold both to record)
  • Instant waveform animation with idle shimmer
  • Security hardening and data integrity fixes
  • "Loading model" indicator on first transcription
  • Model stays loaded between recordings for instant use
Version 1.0.0 (Build 103)
  • Instant waveform animation on recording start
  • Subtle idle shimmer shows the recorder is alive during silence
Version 1.0.0 (Build 102)
  • Shows "Loading model" indicator during first transcription
  • Model stays loaded between recordings for instant subsequent use
  • Security hardening and data integrity fixes
Version 1.0.0 (Build 101)
  • Security: license validation hardened, trial tamper detection
  • Data integrity: orphan cleanup no longer deletes referenced audio files
  • Faster: whisper model stays loaded between recordings
  • Fixed: transcription state recovery on unexpected errors
  • Fixed: SwiftData corruption recovery path
Version 1.0.0 (Build 100)
  • Consolidated sidebar: License, Permissions, and About moved into Settings
  • History: date groupings, TXT/CSV export, Share sheet
  • Thumbs up/down quality feedback on transcriptions
  • Menu bar: last transcription preview with Copy
  • Confetti celebrations on transcription milestones
  • Recording timer color shift at 60s and 2min
Version 1.0.0 (Build 99)
  • RAM-based model gating - prevents downloading models too large for your Mac
  • In-app update banner with one-click relaunch
  • Shows "Loading model" during first-use model load instead of "Transcribing"
  • Fix Settings tab causing recording lag
Version 1.0.0 (Build 98)
  • In-app update banner with one-click relaunch
  • Fix Settings tab causing recording lag
Version 1.0.0 (Build 97)
  • Fix auto-learn corrections not detecting edits after single recording
  • Fix approval toast stealing keyboard focus from other apps
  • Fix secondary hotkey becoming invisible and impossible to remove
  • Dictionary view: collapsible sections grouped by source and pack
Version 1.0.0 (Build 96)
  • Fix data store path for reliable persistence across builds
  • Fix dictionary migration for upgrading users
Version 1.0.0 (Build 95)
  • Fix paste not working in terminal apps on first try
Version 1.0.0 (Build 94)
  • Fix static noise on first recording after sleep
Version 1.0.0 (Build 93)
  • Dictionary learning now asks before saving - approve or reject each correction
  • Fix missing capitalization after sentence-ending punctuation
Version 1.0.0 (Build 92)
  • Fix repeated words at end of transcription from trailing silence
Version 1.0.0 (Build 91)
  • Remove ellipsis artifacts from mid-speech pauses
Version 1.0.0 (Build 90)
  • Stability improvements for long-running background sessions
  • Auto-restart on unexpected termination
Version 1.0.0 (Build 89)
  • Fix background crash from re-entrant constraint updates in menu bar icon
Version 1.0.0 (Build 88)
  • Copy confirmation feedback on dashboard transcription card
Version 1.0.0 (Build 87)
  • Fix duplicate window appearing on Dock click or Cmd+N
Version 1.0.0 (Build 86)
  • Whisper Lite is now the default model for all installs
  • Apple Speech available as optional alternative on macOS 26
Version 1.0.0 (Build 85)
  • Apple Speech auto-selected on macOS 26, Whisper Lite on older Macs
  • Friendly model names: Whisper Large, Whisper Fast, Whisper Lite
  • Honest model descriptions highlighting trade-offs
  • Upgrade nudge for pre-macOS 26 users in AI Models
  • Fix missing menu bar icon
Version 1.0.0 (Build 84)
  • Fix background crash caused by re-entrant constraint updates
  • Remove unnecessary permission polling timers
Version 1.0.0 (Build 83)
  • Add Open at Login toggle in Settings > Advanced
Version 1.0.0 (Build 82)
  • Fix background crash caused by re-entrant constraint updates
  • Globe key labels now display correctly in hotkey picker
  • Interactive hotkey picker in onboarding (choose your recording key)
Version 1.0.0 (Build 81)
  • Fix runaway CPU usage caused by media playback monitor EOF spin loop
Version 1.0.0 (Build 80)
  • Fix missing app icon in release builds
  • Mic standby off by default, optional "Keep Microphone Ready" toggle
Version 1.0.0 (Build 79)
  • Mic standby off by default - no more orange indicator in menu bar
  • Optional "Keep Microphone Ready" toggle in Settings for zero-latency start
Version 1.0.0 (Build 78)
  • Auto-learned corrections shown in "Most Recent" dictionary section
  • Permission reset buttons (revoke and re-grant from within the app)
  • Reduced whisper hallucinations (trailing silence trimming, tuned thresholds)
Version 1.0.0 (Build 77)
  • 8 industry dictionary starter packs (535 entries)
  • In-app feedback popover replaces Google Forms
  • Onboarding dictionary personalization step
Version 1.0.0 (Build 76)
  • Fix trial banner showing alongside active license
  • Auto-learn correction timer extended to 45s
Version 1.0.0 (Build 75)
  • Zero-latency recording start with lookback ring buffer
  • Sidebar waveform meter animates during recording
  • Version badge polish
Version 1.0.0 (Build 74)
  • Dedicated License tab in sidebar for easy activation
Version 1.0.0 (Build 73)
  • Error notifications for download failures, transcription errors, and permission issues
Version 1.0.0 (Build 72)
  • Check for Updates in app menu
Version 1.0.0 (Build 71)
  • Fix microphone permissions in signed builds