KeySpeech 1.4.1 lets you set the voice keyboard's height and fixes the keyboard on iOS 27. • Voice keyboard height: Settings › Keyboard, from half to the standard height. When it is low, delete and @ sit side by side above return. • iOS 27: the keyboard is the system keyboard's height again, with no extra panel above it. • Full keyboard: give the language key a color of its own — Settings › Keyboard › Layouts. • The full keyboard has a space bar in a browser's address bar now. • History: the buttons under a dictation no longer break their words.

KeySpeech 1.4 adds an offline mode, voice recordings in History and a Dictionary that learns from your dictations. • Offline mode: one switch at the bottom of the keyboard's provider menu, or in Settings › Keyboard. Dictation and new notes are recognized on your iPhone, without AI, and your templates stay as they are. • Network gone mid-dictation? KeySpeech recognizes the recording on your iPhone instead, if a speech model is on it. If not, the dictation waits in History with its recording. • Cancel a dictation: tap the cross in the corner of the Record key or swipe left. Nothing is inserted. • Voice recordings in History: keep the audio of your last 20, 50 or 100 dictations, or none. Play one back or recognize it again with another engine. • Find words for your Dictionary: your AI provider reads your latest dictations and picks out words the recognizer gets wrong. Untick the ones you don't want, tap a spelling to correct it and add the rest at once. • Notes now have four tabs: Original, Text, Speakers and Summaries. With AI processing on, the speakers' lines are tidied up too. • Choose which models the keyboard's model menu offers: Settings › Keyboard › Models in the menu. • Set one provider and model for all summary templates at once. • "Where you dictate" on the Dashboard now lists apps only, not notes.

KeySpeech 1.3 adds a second keyboard and redesigns the whole app. • Full keyboard: letters laid out like the iPhone's own, with dictation in a bar above them. Eight layouts, accents, emoji with search and a space-bar trackpad for the cursor. Pick it in KeySpeech's Settings › Keyboard. • Templates with their own AI: every dictation and summary template keeps its own provider and model. Chips on both keyboards switch the template, provider and model. • Redesigned app. The Dashboard shows the time you've saved, your speaking pace, where you dictate and your daily rhythm. History is grouped by day and shows the app each dictation went into. Notes get a player with a waveform colored by speaker, and Text, Transcript and Summaries tabs. • Redo a note: recognize it again, separate the speakers again, or process it with another template, provider or model. • Import audio and video files from Files into Notes. • Speaker separation through OpenAI, with your own key. KeySpeech asks before uploading a recording. • Each speech engine keeps its own language, with many more languages to choose from. • Microphone standby from 5 minutes to 2 hours, with a real pause, and a choice of how long History is kept.

Speakli is now KeySpeech: the same app, with your notes and settings, under a name that's easier to find. • AI on your iPhone (optional): download the Qwen3.5 4B model once (about 3 GB), and summaries, the cleanup of new notes and their titles run on the iPhone itself, with no API key and nothing sent anywhere. Needs an iPhone with 12 GB of memory, such as iPhone 17 Pro or 17 Pro Max, and works while KeySpeech is open. • Titles by AI: a new note gets a title from its whole text instead of its first words. Titles you write yourself are never changed. • Smaller fixes and improvements.

Speakli 1.1 turns Notes into a meeting recorder. • Record up to 4 hours per note with your iPhone locked. A call or Siri pauses the recording, and it resumes afterwards. • A Live Activity on the Lock Screen and in the Dynamic Island, with a Stop button that works without unlocking. • Transcripts by speaker, separated on your iPhone with an optional ~22 MB model. Name each voice once, and Speakli can suggest who's speaking in later notes; you confirm. • Summaries from your own AI provider: 12 templates, including Meeting summary, Action items and Formal minutes, or your own request. Tap a time in a summary to hear that moment. • Export transcripts as TXT, Markdown, SRT, WebVTT, RTF, PDF or Word, plus summaries, a full report and the audio file. • Transcribe again with another engine, and play the recording from any paragraph. • New Dictionary: names, brands and terms spelled your way, with "Also heard as" variants. Add words straight from History. • Adjustable size for the live text on the Record key.