After dictating with the voice keyboard, tap the text to copy all of it. The keyboard confirms with "Copied" in its header, and for your first few dictations it reminds you that the text can be tapped. The voice keyboard no longer quits on its own after you switch between apps or fields a few times. It was holding on to every copy of itself it had ever shown until iOS closed it. Typing feels closer to the system keyboard. The Chinese candidate bar keeps up with fast typing, a tap in the gap between two keys types the nearer one, keys light up the moment you touch them and click if Keyboard Clicks is on, holding delete speeds up and then deletes whole words, and English sentences start with a capital letter. Dictating again within 30 seconds of the last dictation starts right where you are, without opening Parley. A single dictation can now run for 10 minutes instead of 2, the keyboard counts down its last 30 seconds, and whatever you said before the limit is typed. If the connection drops, the words you already said are typed instead of lost. When a transcript comes back empty or cut short, a recording keeps failing to sync, or the microphone kept dropping, Parley offers to send us the diagnostics in one tap, never your recordings or transcripts. You can also report a problem from Settings, and crash reports are sent automatically; you can turn that off in Settings.

The Parley keyboard looks right in dark mode, in every app. In apps that follow the system's dark mode — Claude and LINE among them — the keys were drawn white with black letters on a dark keyboard. They now match the keyboard around them, and change with it when you switch dark mode on or off while typing. 注音 types Chinese punctuation. 「，」 and 「。」 sit on either side of the space bar, and the 123 and #+= keys on the 注音 keyboard now bring up full-width marks — 、？！：；「」（）…～ and more. Tapping space twice ends a sentence with 「。」. Tap ⌄ at the end of the candidate bar to see every candidate at once, in a grid where the keys were. Tap one to type it, or ⌃ to go back to the keys. More candidates stay in view while you type a long phrase: the reading above the keys now shows its last two syllables instead of all of them, so the candidates keep most of the bar. The keyboard is lighter and quicker. Typing on the 注音 and English keyboards now redraws only what changed — the bar of candidates above the keys — instead of every key on every tap, so the keyboard keeps up when you type fast. It also holds less in memory. Its word lists are never loaded twice at once, and the ones you are not using are let go when your phone runs short, which makes the keyboard much less likely to close unexpectedly in the middle of a sentence. One wrong 注音 symbol no longer empties the candidate bar. Mix up two symbols that sound alike — n and l, a retroflex and its flat twin, a front and a back nasal — or catch the key next to the one you meant, and the 注音 keyboard still offers the word you were after, right behind the words that match exactly what you typed, so typing correctly works just as it did. A syllable no character is read as now commits its closest character on return instead of the raw 注音. Tones still have to be right.

Dictating twice in a row stays where you are. After a dictation, Parley now keeps the microphone for 30 seconds, even with "Keep the microphone ready" off. Tap the mic again within that time and it records right in the app you are typing in, instead of bringing Parley forward a second time. iOS shows the orange microphone dot for those 30 seconds, because Parley really is holding the microphone; nothing is recorded, transcribed, or sent until you tap. A dictation you start right after stopping the last one shows only its own words. The previous one can no longer flash into it for a moment or be typed a second time. Swipe between Voice, English and 注音 from anywhere on the keyboard, including the empty space on the voice pane, not only from the record button. While you dictate, delete, return and @ now stay in place, dimmed, instead of disappearing, so the voice pane no longer looks like it emptied out.

The English keyboard suggests words as you type. Start a word and the likeliest endings line up above the keys; tap one and it lands with a space after it. Nothing is ever rewritten on its own: space and punctuation type exactly what you typed, and a suggestion only goes in when you tap it. Capitalisation follows what you started. The voice pane is quieter while you speak. Delete, return and @ step aside for as long as the microphone is open — nothing has landed in the field yet, so there was nothing for them to act on — and come back the moment the words do. And once voice typing is set up, the record button is a microphone. It used to show an "opens Parley" arrow whenever the app was not already awake in the background. The first tap after a while may still bring Parley forward once, but that is how it starts, not something the button needs to warn about.

Voice typing now says so on the Lock Screen, and the record button listens. Swipe the keyboard away mid-sentence and Parley is still listening — which until now was something you could only find out by bringing the keyboard back. Now it moves to the Dynamic Island and the Lock Screen, with the time running, so you can see it without unlocking. It also covers the quiet state that was hardest to explain: when you have asked Parley to keep the microphone ready between dictations, the card says the microphone is open, says plainly that nothing is being recorded, counts down how long is left, and lets you end it in one tap. None of your words appear on the card, in either state. A Lock Screen is a public surface, and what you are dictating stays inside the app. The record button follows your voice. It used to pulse on a loop whether or not anyone was speaking; now it swells with what you actually say and rings ripple out from it, and in silence it is completely still. And you can feel more of it. Opening the microphone is a rising two-beat rather than a single tap that got lost against the keyclick. Swiping the keyboard away while a dictation is still running has its own falling one, so you know the microphone stayed behind without looking. Stopping a recording has a beat that comes to rest. And if iOS takes the microphone — its own dictation key sits just below this keyboard, and it is easy to hit by mistake — you feel that too, instead of talking on into nothing.

Recordings can be played back on the phone. Meetings recorded here now stay on the phone (Settings → Keep audio on this phone, on by default), and a recording made elsewhere can be downloaded from its row, its long-press menu, or the transcript screen. Open a recording and a player sits under the title: the whole meeting as a waveform, drag it to scrub — lift your finger higher while dragging for finer control — a speed control from 0.75× to 2×, and hold the edge of the transcript to play at 2× while you hold. The transcript follows the audio and marks who is talking, and tapping a timecode — or anywhere in a turn's text — jumps the audio there. Search inside a transcript. Tap the magnifying glass on a recording, type, and every match is highlighted in place, with a count and next/previous arrows at the bottom; tapping a highlighted match plays from that turn. The library search still covers titles and snippets — this one covers the words. A recording can be transcribed again. When the transcript came up thin or wrong — the relay dropped out, the room was loud — open the recording, tap ⋯ and choose Re-transcribe. The whole audio goes through transcription once more and the words are swapped in when it comes back; the speaker names, highlights and folder you already had stay exactly as they were. It uses your account's transcription hours the same as a new meeting would, and each recording can be re-run three times. The app looks different. The page is white, the text is ink, and blue is reserved for what is happening right now: the person speaking, the live waveform, the thing you can tap. Speakers are Speaker A, B and C instead of numbers. While recording there is a real waveform — calm, rounded and pale — and a Discard action for a recording that should not have started, so a mis-tap no longer leaves a two-second file in the library. After a recording, the suggested name and folder are one card and one decision: save as suggested, adjust, or skip. In the library, swipe across the header to move between folders, and the import button no longer wears a download arrow. The keyboard admits when the microphone is gone. If iOS's own dictation takes the microphone mid-session, the voice pane now says so and a tap starts a fresh session, instead of showing ⏹ over silence until Parley was force-quit. Since 1.7, also: the phone names and files its own recordings; an audio file from anywhere can be imported and transcribed here; and a meeting whose live transcript came up short is transcribed again from the audio.

Dictation now arrives written, not transcribed. When a session ends, Parley makes one cleanup pass over the text before the keyboard inserts it: filler words go, punctuation lands where you meant it, and a long stretch breaks into paragraphs. What you said in a hurry reads like something you typed. It stays in your language. Traditional Chinese comes back Traditional — a result that drifted into Simplified is discarded rather than pasted — and the words you have taught the personal dictionary are protected through the pass and reapplied after it, so a correction you made by hand always wins. The pass costs a moment at the end of a dictation, and it can only help: on a timeout, with no network, signed out, or if the result comes back wrong in any way, the keyboard pastes the raw transcript exactly as it did before. Turn it off in Settings → Voice keyboard → Polish with AI. The keyboard now answers you. Starting a dictation gives one solid tap, and the words landing in your field give the two-beat pattern iOS uses for anything that finished — so you can start speaking and know it took, and look up only when you feel it arrive. Both need Full Access, which is the only way iOS lets a keyboard play haptics at all; without it nothing else changes.

The keyboard grew up in this release: - New Bopomofo (Zhuyin) keyboard - swipe between the voice, English, and Traditional Chinese panes, and choose which keyboards you carry in Settings. - The voice pane now says exactly what a tap will do: when the microphone window is open you dictate right where you are; when Parley still needs setting up, the button honestly opens the app instead of pretending to listen. - Dictation now lands in your document in one piece when the session ends, instead of streaming in word by word - cleaner to read, easier to edit. - Personal dictionary: fix a word right after dictating and Parley learns your spelling for next time. Review, add, or remove entries in Settings > Personal dictionary.

Dictation can stop leaving the app you are typing in. Turn on a microphone window in Settings — five minutes, fifteen, or an hour — and after a dictation Parley keeps the microphone for that long, so the next tap on the keyboard's mic types where you already are instead of opening Parley. iOS shows the orange microphone dot for the whole window, because Parley really is holding the microphone: nothing is recorded, transcribed or sent until you tap the mic, and sound that arrives before then is thrown away as it comes in. It is off until you turn it on, every window ends on its own, and you can end one early from Settings, from the record screen, or from the keyboard. Losing the connection is now a pause rather than the end. A network drop mid-sentence used to finish the recording and show an error; the transcript now reconnects on its own, keeps the audio from the gap instead of discarding it, and only gives up once it really cannot get back. The voice half of the keyboard was drawn like a keyboard, and it isn't one: nothing on it types a letter. It is now a round record button in the middle of the pane, with delete and return as small round keys under your right thumb and @ in the bottom-left corner. Return used to be the widest key on the keyboard and the least pressed one. You can watch the words arrive. The sentence being transcribed sits above the button, with the part already placed in your text in a lighter grey — dictation used to look like words that appeared and then vanished. The globe key now appears only where iOS doesn't already draw one. 1.3 put it on every device; on iPhone X and later the system already places its own globe and dictation keys below any keyboard, this one included, so ours was a second copy taking up a slot. On the phones without that strip it is still there, and the system Chinese keyboard is still one tap away. Swiping between voice and typing follows your finger instead of jumping when you let go, and the keyboard no longer changes height as you cross between them. Recording holds on to the microphone better. A recording survives leaving the foreground, one tap on record means one recording and not two, and the microphone permission prompt now survives the hand-off from the keyboard — coming back with permission granted starts dictation instead of stranding you.

The keyboard is a keyboard now. Next to voice it has a full English layout — letters, numbers, symbols, hold-to-repeat delete — and the keys you reach for in the middle of dictating without leaving the mic: return, @, and delete. Swipe left or right across the keyboard to move between the two. The globe key is on every device now, so the system Chinese keyboard is one tap away and holding it opens the picker. Parley does not ship its own Bopomofo layout; the way out of this keyboard is no longer hidden on some phones. The keyboard also lines up with the screen. It used to paint its own background over the system's, which left a seam along the bottom edge and a top corner that did not match — it now sits in the system's own input view, at the system's own key metrics. Transcript text can be selected and copied, while it is still being transcribed and not only after. Copy one line from its context menu, or the whole transcript from the toolbar. And the app looks like Pathors: brand blue, a lighter page, and the type the rest of Pathors is set in.

Voice typing now happens right where you are. Tap the mic on the Parley keyboard and your words stream in at the cursor — and after you stop, tapping the mic again keeps you in your app instead of bouncing back to Parley. Returning to your app no longer loses the transcript, Action Button dictation is more reliable, and the keyboard setup in Settings is simpler. Plus the usual fixes.