Speech2TextAI
Free trial Buy for £9.95

Changelog

Every purchase includes 1 year of free updates — the app checks daily and updates itself with one click. Updates never touch your settings: your dictionary, snippets, styles and history all carry over.

v1.22.1 (28 October 2026)

  • No more silent hotkey in admin windows. While a window running as administrator is in front (for example a terminal opened with “Run as administrator”), Windows hides your keyboard from normal apps, so holding the hotkey there did nothing. Speech2TextAI now tells you when this happens, once per app, and offers Restart as administrator in the tray menu so you can dictate into admin windows too.

v1.22.0 (22 October 2026)

  • Styles no longer slow your dictation down. Speech2TextAI servers now transcribe every dictation first, in well under a second. If you’ve set a style for the app you’re typing into (say, professional for email), it’s applied straight afterwards and adds a couple of seconds at most. If that step is ever slow, your plain transcript is typed immediately instead of making you wait.

v1.21.1 (20 October 2026)

  • Much faster dictation if you picked your own language as the output. Setting “Type it in” to the same language you speak (for example English (UK) → English (UK)) was treated as a translation, which sent every dictation the slow way. It now goes through Speech2TextAI servers as normal — typically under a second.
  • Clearer settings. The Styles and Language tabs now say up front that styles and translation take a few seconds longer, so you can choose speed or styling.

v1.21.0 (1 October 2026)

  • A proper tray panel. Left-click the Speech2TextAI icon by the clock and you get a neat card showing whether it’s listening, a reminder of the hotkey, and all the actions: pause, copy or paste the last transcript, History, Transcribe a file, Settings, Help and Quit. Right-click still shows the plain menu.
  • It tells you where it lives. The first time it starts, a balloon says it’s running in the tray and how to dictate. Starting it again while it’s already running now pops the panel instead of doing nothing.
  • No more silent exits. If no microphone can be found, you now get a clear message with a Retry button instead of the app quietly closing.
  • History inside Settings. Everything you’ve dictated is now also a tab in Settings, so you don’t have to go hunting for it.

v1.20.2 (29 September 2026)

  • Copy lots of history at once. In History, select several entries with Shift or Ctrl (click, or Shift with the arrow keys, Home and End), or press Ctrl+A for everything. “Copy selected” now copies every selected entry, oldest first, instead of just one. Ctrl+C works too.
  • Clearer wording. The Transcribe a file window no longer promises speaker labels, which need the optional Gemini key.

v1.20.1 — 29 September 2026

  • No more missing spaces when you dictate back-to-back. If you held the hotkey for your next dictation while the previous one was still being typed, the spaces could disappear (“likethis”). The app now pauses typing while any Ctrl, Win, Alt or Shift key is held and carries on the moment you let go, so everything lands in order, correctly spaced.

v1.20.0 — 28 September 2026

  • No Google account needed any more. Speech2TextAI now transcribes on its own Speech2TextAI servers by default — install it, hold the hotkey and talk. There’s no API key to create.
  • Much faster and far more reliable. A typical dictation now comes back in well under a second, and it no longer depends on Google’s busy periods, which caused slow or failed dictations through September.
  • Your Gemini key becomes an optional backup. If you already have one, keep it: it’s used automatically if the servers are ever unavailable, and it powers Styles and translation. You can choose the engine in Settings → General → Transcription.
  • File transcription on the servers too. Drop in a recording up to an hour long (95 MB) and it’s transcribed without a key.
  • Private by design. Audio is processed on Speech2TextAI’s UK servers and never stored.

v1.19.1 — 23 September 2026

  • Keeps working through Google outages. Google’s Flash-Lite models have been overloaded on and off since 22 September, returning “high demand” errors for hours at a time. The app now falls back to a different Gemini model that stays available, so dictation carries on instead of failing.
  • Updates always fetch the newest build. The in-app updater could receive a cached copy of the previous version from Google’s and Cloudflare’s caches. Each release is now published under its own file name and the app asks for a fresh copy.

v1.19.0 — 22 September 2026

  • A fresh, cleaner look. Every window has been redrawn: rounded buttons and fields, proper dropdowns, grouped settings cards, a formatted Help page and a new Analytics view with daily stats and a 7-day chart, in both light and dark themes. Switching theme now updates Settings instantly.
  • No more false “check your internet” when Google is busy. When Google’s service is slow or overloaded, saving your API key used to blame your connection. It now tries every model, tells you honestly when Google itself is busy, and keeps your key.
  • Slow microphones no longer corrupt later dictations. A mic that took a few seconds to wake up could leave a hidden recording running in the background and mix stray audio into everything you dictated afterwards. Fixed.
  • Feedback window fixed. The Send button could sit below the bottom of the window on some screens, so feedback couldn’t be sent. The buttons are now always visible.
  • Smaller fixes. A failed manual update check now shows what went wrong instead of nothing; the trial-ended window no longer clips its buttons; entering a licence key during a trial takes effect straight away.

v1.18.1 — 27 August 2026

  • Faster, steadier dictation. When Google is slow to answer, the app now races a second request alongside the first and takes whichever comes back sooner — the occasional 10–30 second wait is gone, and a slow spell no longer leaves you stuck on a slower model until you restart.
  • Works with admin windows. Dictating into a window running as administrator used to do nothing at all (Windows blocks normal apps from typing there). Your text is now copied to the clipboard with a note to press Ctrl+V.
  • Update check fixed. “Check now” no longer reports that it can’t reach the update server.
  • Microphone problems are explained, not silent. If your mic is muted, asleep or unplugged you get a clear message the moment you press the hotkey — and the app never freezes waiting for it.
  • A cleaner look. Refreshed Settings with tidier tabs, proper toggle switches and grouped sections, in both light and dark themes.

v1.17.1 — 19 July 2026

  • The error sound is now three beeps (was two), so it’s easier to tell apart from the others.

v1.17.0 — 18 July 2026

  • Scrollable settings. Every settings tab now scrolls, so nothing gets cut off — no more resizing the window to reach the options at the bottom.
  • Language gets its own tab, with a built-in tester — set your languages and record a quick test to hear the translation, right there.

v1.16.0 — 18 July 2026

  • A sleeker, cleaner interface. The whole app has been redesigned with a modern, refined look — clearer tabs, better spacing and a calmer colour scheme.
  • Light & dark themes. Choose whichever suits you in Settings → General → Theme.

v1.15.0 — 18 July 2026

  • Hands-free mode. Turn it on in Settings and just tap the hotkey to start and tap again (or press Esc) to stop — no need to hold the keys down. Perfect for long dictations.
  • Live batches for long dictations. Choose to have your words typed out every 15/30/45 seconds or every 1/5/10 minutes, so you watch the text arrive in real time and never lose a long session if something interrupts it.
  • Speak any language — and translate on the fly. Pick the language you speak and the language you want it typed in. Dictate in Spanish and get English, speak English and get French — the translation is done by the same AI, no other service involved. 30+ languages.

v1.14.0 — 18 July 2026

  • Choose your own hotkey. Ctrl+Win clashing with a Windows shortcut? Pick a different combo in Settings — Ctrl+Alt, Alt+Win, Ctrl+Shift or Right Ctrl.
  • On-screen status. A small floating pill shows “Listening…” then “Transcribing…” so you always know what the app is doing (turn it off in Settings if you prefer just the beeps).
  • Your API key is now encrypted on disk using Windows’ built-in protection, so it can’t be read straight out of the settings file.
  • Stronger licensing under the hood: the app no longer carries any secret that could be used to forge licence keys — validation is handled securely by the server with signed activation.

v1.13.0 — 18 July 2026

  • Your text goes where you meant it. If you switch windows while a dictation is still processing, it’s now copied to the clipboard (with a heads-up) instead of being typed into the wrong app. Quick back-to-back dictations always arrive in the order you spoke them.
  • Copy / Paste last transcript added to the tray menu — your last dictation is always one click away, even if something went wrong.
  • No more accidental half-sent messages: multi-line dictations (lists, paragraphs) are pasted as one block, so a stray Enter can’t fire off a partial message in Slack, Teams or chat.
  • Clearer feedback: friendly notifications when no speech is heard or the microphone isn’t available — instead of silence.
  • Custom-word fixes are smarter (a rule for “art” no longer touches “start”), transcripts are kept out of the log file by default for privacy, and licences never lock you out during a server hiccup.

v1.12.0 — 18 July 2026

  • Name the speakers: after transcribing a recording, a “Who’s who?” row lets you replace Speaker 1, Speaker 2… with real names throughout the transcript.
  • Email it to yourself: one click sends the finished transcript to your inbox (up to 10 a day).
  • Developer term pack: one button preloads your custom dictionary with 50+ dev terms and common mis-hearing fixes — GitHub, JSON, kubectl, nginx, “j son” → JSON, “engine x” → nginx and friends. Edit or delete anything you don’t want.

v1.11.1 — 18 July 2026

  • Fixed: silent dictations no longer invent text. If you held Ctrl + Win without speaking, the AI could hallucinate a plausible-looking sentence. A new voice-activity check now detects silence, fan hum and keyboard noise locally and simply ignores the press — nothing is sent, nothing is typed.
  • Bonus: silent presses no longer use your API quota at all.

v1.11.0 — 18 July 2026

  • Transcribe your existing recordings: new “Transcribe a file” window (tray menu) — drag & drop a voice memo, meeting or interview, or click to pick one. You get a clean, punctuated transcript with speaker labels, a Copy button and Save-as-.txt.
  • Supports wav, mp3, m4a, aac, ogg, flac, aiff and mp4 audio. Small files transcribe directly; big files (up to 500 MB) upload to Google automatically first.
  • Your custom dictionary applies to file transcripts too, and they’re kept in History.

v1.10.0 — 16 July 2026

  • Tabbed Settings: everything reorganised into clean tabs — General, Dictionary, Styles, Test lab, Sounds, About.
  • Test lab: record a test dictation and see the result on screen instead of it being typed into an app — perfect for trying out custom words, snippets and styles before you rely on them.
  • Free-updates countdown: Settings now shows exactly how long your free updates run (1 year from first activation).
  • “What’s new” button linking to this changelog.

v1.9.3 — 16 July 2026

  • Per-app styles, point-and-click: pick a common app (Terminal, VS Code, Word, Outlook, Slack, Teams…) from a dropdown and a tone is suggested for it — plus a one-click “Programmer preset” for all dev apps.
  • Customisable beeps: choose a pitch for each of the three beeps (recording / transcribing / error) or turn any of them off, with a Test button to hear it.

v1.9.2 — 16 July 2026

  • Fixed a false warning for some Google Workspace API keys that actually work fine.
  • Genuinely blocked work accounts now get a clickable link straight to the Google Cloud page where IT can enable the API.

v1.9.0 — 15 July 2026

  • Free 28-day trial: the full app, no card, no licence key needed.
  • Update notifications now appear as a quiet tray balloon instead of a pop-up window.
  • Clearer setup guidance for work (Google Workspace) accounts.

v1.8.0 — 14 July 2026

  • Licence covers 2 computers, with a “Deactivate this computer” button to move machines.

v1.7.0 — 14 July 2026

  • Styles: a different tone per app — professional in Outlook, casual in chat, technical in your editor.
  • Type-out is now the default insert method (more reliable than the clipboard in fussy apps).
  • In-app feedback form (Settings → Feedback).

v1.6.0 — 13 July 2026

  • Wispr-style intelligence: mid-sentence corrections (“2 — actually 3” types 3), spoken lists become numbered lists, natural punctuation.
  • Snippets: say a cue (“my sign-off”), get your full saved text.
  • Choice of insert method (type out / clipboard) and a beeps on/off toggle.

v1.5.0 — 13 July 2026

  • Custom dictionary: teach it names, jargon and corrections (“wrong => right”).

v1.4.x — 12–13 July 2026

  • One-click in-app updater.
  • Automatic fallback across Gemini models during Google outages — dictation keeps working.
  • Step-by-step free API key instructions in the setup window.

v1.3 & earlier — 11–12 July 2026

  • Dark console theme across every window.
  • History viewer, analytics (words/day, time saved), microphone picker, licence keys.
  • The core: hold Ctrl + Win, speak, release — clean text at your cursor, in about a second.