Release Notes

Diction 8.0

• Brand-new autocorrect engine, the first real one in Diction. It quietly fixes common mistakes as you type, with no setup required.
• The QWERTY keyboard layout has been updated: more compact, more responsive, and easier to use. Suggestion chips now sit inline next to the mic button, and the dismiss key lives in the keyboard row itself.
• Suggestions are significantly more reliable. They now track context correctly after backspace and edits, and tapping them works consistently wherever you are.
• When you have text selected and speak an edit instruction, the live streaming preview no longer overwrites your selection mid-way. The replacement arrives as a single, clean swap.
• Tap the mic inside any transcript to record more and insert it at the cursor. In Live mode, the detail view opens right away so text streams in as you speak.
• Added a numpad for number fields. Tap into a phone number, PIN, or numeric field and the keyboard adapts automatically.

Diction 7.0

• A built-in QWERTY keyboard. When you can't talk out loud, just type without leaving Diction. Autocorrect, word suggestions, long-press accents, key popups, caps lock, double-space-to-period, backspace acceleration, and the feel of a native keyboard.
• Diction Live (Beta). Your words now appear as you speak them, in real time, instead of only after you stop.
• Automatic language detection. Switch between languages a lot? Diction can now work out which one you're speaking. No need to set it each time. Optional, off by default.
• Smarter capitalization. Dictation now respects the field you're writing in, so email addresses, code, and similar text don't get wrongly capitalized.
• A cleaner mode switcher. Picking Diction One, Live, an on-device model, or your own server is now a dedicated one-tap screen, with your current plan shown at a glance.
• Record and transcribe right inside the app. A new mic button on the home screen lets you capture speech without the keyboard, then copy the text or send it straight to any app with the share button.
• More reliable in the moment. Better retry and fallback when a transcription fails, sturdier live-streaming connections, and a refined suggestion strip that stays out of your way.
• The keyboard's default key row now includes a period key, and choosing an on-device model starts downloading right away with no extra confirmation step.
• Fully localized across all 13 languages, with faster recording start, proper light and dark keyboard theming, and lots of smaller polish and stability fixes.

Diction 6.0

• Customizable keyboard row. Tap any key to change what it does, rename it, resize it. Add new keys, drag to reorder, delete what you don't need. The whole row is yours now.
• Per-key visibility. Toggle "only show in edit mode" so custom keys stay hidden while you're dictating and appear when you select text. No more layout jumping.
• Tap-to-fix suggestions. Select a word and three smart alternatives appear above the keyboard: spelling fixes, grammar corrections, or just a better way to say it.
• Save to My Words. A new key that teaches Diction your names and jargon with one tap. Custom words feed into transcription for better accuracy going forward.
• Undo and redo. Tap to reverse your last dictation, tap again to bring it back. Works for edits too, not just transcriptions.
• Diction One now includes both cloud and on-device. Subscribers get the flagship cloud model and the best offline model, so you're covered with or without a connection.
• More on-device models. Whisper Small and Nvidia Parakeet join the lineup, giving you more choices for different languages and accuracy needs.
• Redesigned home screen. Three clear tiers: Diction, On-Device, and Self-Hosted, each remembering your selection when you switch between them.
• Continued improvements to speak-to-replace. Better casing, punctuation matching, and context awareness when editing selected text.
• Refreshed onboarding with illustrated steps.
• Self-hosted: the gateway now speaks the OpenAI speech API, so any compatible SDK works as a drop-in.
• Improved reliability, bug fixes, and various UI polish.

Diction 5.0

April 2026

• Profile lets you tell Diction who you are and how you write. AI Companion uses it every time, alongside whatever Tone Preset you have set for the app you are in.
• Diction now figures out what you want from where your cursor is. Cursor in empty space, it dictates. Cursor inside a sentence, it inserts. Text selected, it edits or replaces. No more switching modes.
• New "After dictation" mic timeout releases the mic the moment your transcription comes back. Plus 10 and 30 second steps for short bursts. If you use Bluetooth headphones, music and podcasts no longer drop into lower-quality voice-call audio while you are holding the mic.
• Long dictations come back complete. Long messages, long notes, even full meetings work end to end without truncation.
• AI Companion is tuned to sound more human. Less of the em-dash-everywhere, faintly corporate cadence that gives away machine-cleaned text.
• Your words come through as you said them, including profanity. No automatic censoring.
• Redesigned insights dashboard with a daily time-saved card and a words-per-minute card showing how much faster you are than typing.
• Double-tap space now inserts a period and a space, matching the standard iOS keyboard convention.
• Smoother onboarding with steadier layout when the keyboard opens, plus a small pulse on the active step.

Diction 4.0

April 2026

• Speak to Edit. Select any text, say what you want changed, and it's done. Works for simple replacements and editing instructions like "translate to Czech" or "make this shorter."
• Your custom words now improve transcription accuracy directly. Names and jargon get recognized correctly even without AI Companion.
• Dictate for as long as you need. Improved reliability for long recordings, no more cut-off transcripts.
• Profile lets you tell Diction who you are and how you write, so AI Companion matches your style.
• New guided onboarding walks you through setup step by step instead of throwing dialogs at you on first launch.
• Improved on-device model setup. Smoother download, faster preparation, automatically ready when done.
• The mic no longer activates when you open the app manually. Orange dot only when you're actually dictating.
• Improved AI Companion accuracy across apps.
• Various UI polish across the keyboard, history, tones, and settings.

Diction 3.0

March 25, 2026

• AI Companion now understands where you are in your document. Dictate into the middle of a sentence and it inserts correctly. Select text and tell Diction what to do with it, and it rewrites the selection in place.
• A small pulse now lights up in the action bar when Diction hears your voice. You always know exactly when it is listening.
• Added a dominant hand setting. If you prefer left-handed use, flip the keyboard layout so the controls are on your side.
• Redesigned History with a tabbed view separating recent and all transcriptions. Search works across all of them, and tapping any entry copies it instantly.
• Carefully revisited what happens when a transcription fails. We now save your audio automatically and show a retry button on the keyboard so you never have to say it twice.
• Added a dedicated Keyboard Preferences screen. Easier to find keyboard settings, and auto-detect language is now on by default.
• Added a clear explanation screen for when Full Access is missing. No more silent failures if the keyboard is not fully set up.
• Fixed capitalization and spacing when dictating into the middle of existing text, and a rare issue where cloud transcriptions could fail silently when multiple keyboard instances were active.

Diction 2.0

March 15, 2026

• AI Companion is now available for cloud mode. After transcription, Diction can optionally clean up your text. It removes filler words, fixes grammar, and polishes the result. Only the transcript is sent for cleanup, never the audio. Off by default, toggle it in Settings.
• Added a setup guide that walks you through keyboard installation and permissions before your first dictation. No more guessing why things are not working.
• Large model downloads now wait for WiFi by default. No surprise data bills from downloading on mobile.
• The cloud subscription is now Diction One, with a redesigned offer screen that makes pricing and what is included much clearer.
• Improved dictation reliability. Fixed the tap-to-reconnect loop, globe key skipping past iOS keyboards, and stale heartbeat issues.
• On-device models now pre-warm after download so your first dictation is fast.
• When something goes wrong, you now see a clear explanation screen instead of a silent failure.
• Added a support screen with troubleshooting steps and a way to reach us directly.
• Various UI polish across the keyboard and settings.

Diction 1.0

March 11, 2026

The first public release. Everything that makes Diction what it is:

• Dictation-only keyboard for iPhone. Tap the mic, speak, and text appears wherever your cursor is in any app. No QWERTY, no distractions.
• Three transcription modes out of the box: on-device for complete offline use, self-hosted to point at your own server, and Diction cloud.
• On-device models in three tiers. The standard model downloads automatically on first launch. Larger models are available for better accuracy.
• Self-hosted mode works with any server running the Whisper API format. One Docker command to get started.
• 99 languages with automatic detection. Speak in your language and Diction figures it out.
• No word limits, no daily caps, no session timeouts. Dictate as much as you want.
• Cloud transcriptions are encrypted before they leave the server. Your audio is processed and immediately discarded.
• The Diction app contains no analytics and no tracking code.
• Configurable idle timeout for hands-free dictation.