NotchLive Release Notes

September 26, 2026

Find the words you saved without opening every session. Search across sessions, saved translations, Clean transcripts, meeting summaries, notes, and dictation history from one keyboard-first palette.

One Search for Your Saved Work

• Press Command-K or choose Search in the sidebar to search your workspace.
• Search everything together or filter by Sessions, Meetings, Notes, or Dictations.
• Matching excerpts help you identify the right result before opening it.

Go Straight to the Match

• Open the matching transcript entry, saved translation, or Clean transcript with matching text highlighted.
• Search the full text of saved content, including notes, rather than relying on titles and previews.

Keep Your Hands on the Keyboard

• Start typing as soon as the search palette opens.
• Move through results with the arrow keys, press Return to open, and Escape to close.

A More Polished Workspace

• The compact search palette follows your window theme.
• Refined sidebar alignment and recent-session hover spacing.
• New interface text is localized across all 20 supported languages.

Workspace search runs locally and requires Pro. Your saved content stays on your Mac. Requires macOS 14 or later. Translation requires macOS 15 or later.

September 14, 2026

Microphone captions feel more responsive and stay readable while you speak. This update also makes it easier to move between meetings without closing NotchLive.

Better Microphone Captions

• Improved microphone capture alongside meeting apps, including Safari-based calls.
• New speech detection for microphone captions helps retain quieter speech and short replies.
• More frequent caption updates, with unstable word guesses held back until they settle.
• Better recovery when microphone devices or audio formats change.

Start Fresh Between Meetings

Use Start fresh… in the live Caption or Translate timeline to clear the current conversation. NotchLive asks first, saves and stops any recording, saves the current meeting summary, and preserves whether captions are running or paused. Saved sessions stay intact.

Small Refinements

• Clearer Apps & Media Language wording alongside Microphone Language.
• A cleaner Notes layout in the captions window.
• Updated localized interface text.

Requires macOS 14 or later.

September 11, 2026

Keep the audio behind your transcripts, bring in recordings you already have, and save translations in their own language tabs. NotchLive 1.12 gives your sessions more room to grow, with a new Home screen to get you started.

Save the Audio, Too

• Recording a session can now save its audio alongside the transcript. Choose Compact for smaller speech recordings—about 22 MB per hour—or Lossless to preserve the captured speech audio without additional compression loss.
• Audio saving is opt-in, off by default, and available with Pro.
• Play recordings inside a session, export the audio, and include it in full-data backups.
• If a saved audio file is removed outside NotchLive, the session explains what is missing and keeps the transcript available.

Bring Your Recordings Into NotchLive

• Import an audio or video file to create a saved transcript with audio playback.
• On supported Apple silicon Macs, imported recordings can also receive local speaker detection for participant review.
• Your original file stays untouched. For video imports, NotchLive keeps the audio rather than storing a second copy of the video.

One Session, Several Languages

• Translate a saved transcript into multiple languages and keep each translation in its own tab.
• The Original remains intact, while every language has its own search, copy, share, and export actions.
• Translation continues when you navigate elsewhere and pauses or resumes around higher-priority live work. Text-only sessions work too; saved audio is not required.

A Clearer Starting Point

• The new Home screen brings together shortcuts for captions, live translation, recording, and importing, plus recent sessions.
• Sessions and their generated Meeting Intelligence results now link to each other.
• A separate, default-off setting controls whether Meeting Intelligence runs automatically in Translate mode. You can also choose Run anyway for the current meeting.

Everyday Refinements

• Improved live-translation request pacing and translation-session recovery while retaining Faster updates and More context.
• Clearer setup guidance and paused-caption states.
• More consistent toolbar hover feedback, larger clickable button areas, and localized controls and messages across all 20 supported interface languages.

Audio saving, file imports, and saved-transcript translation require Pro. Lossless recordings preserve speech-quality mono audio; they are not full-bandwidth stereo recordings. NotchLive requires macOS 14 or later. Translation requires macOS 15 or later. Meeting Intelligence requires macOS 26 and Apple Intelligence availability.

September 8, 2026

Choose between earlier translation updates and waiting for a complete caption. This release also improves language switching and makes the captions window easier to use at smaller sizes.

• Faster updates: Translations update while you speak. This is the default for new users; words may change as more speech arrives, and it may use more battery. Compact view and the notch widget always use this mode.
• More context: In the full captions window, choose to wait for each caption to finish before its translation appears, with fewer changes along the way. A subtle waiting indicator shows when translation is pending, and a small icon identifies the selected mode.
• Smoother language switching: Earlier captions stay readable while their translations wait for a quiet moment or for listening to pause, helping new live captions keep up. Live translations no longer briefly show the original language unless you enable it.
• Smaller windows: The captions window can shrink to 520 points wide, with sidebars that adapt as you resize.

Regular captions and saved-caption translation remain available in either mode. Translation timing depends on the language pair and workload; Faster updates does not mean instant translation.

Requires macOS 14 or later. Translation requires macOS 15 or later.

September 8, 2026

Live captions and saved speech now belong in one app window. The desktop workspace is the main place to listen, browse, review, and manage NotchLive; the notch widget remains available whenever you want a smaller caption view.

Live and Saved Work, Together

• A shared sidebar connects live captions, sessions, meetings, notes, dictations, and Settings.
• Open saved work or change settings without leaving the workspace while an active caption session keeps running.
• Session controls remain within reach across views, and the sidebar can collapse when captions need more room.

A Library Built for Review

• Sessions, meetings, notes, and dictations use consistent cards with titles, useful text excerpts, and clearer metadata.
• Session search covers both raw and cleaned transcript text, with clearer empty and no-result states.
• Sharing, export, and item actions now sit together in cleaner detail headers, leaving more room for the content itself.
• Saved-item navigation keeps its place when you visit Settings and return.

Setup and Appearance That Follow Your Choice

• The first-run experience now begins in the app window and shows clear progress through permissions and model setup.
• Move back to earlier setup steps, try the notch widget, return to the window, or choose Set Up Later and resume from clear guidance.
• Finishing setup opens the caption surface you selected.
• Choose light, dark, or system appearance with Paper, Beige, Charcoal, and Midnight Black shades. Waveforms use clearer colors and subtler glow in light appearances.

Navigation, Localization, and Everyday Polish

• Settings have clearer grouping, aligned navigation labels, and visible upgrade options for free users.
• Saved sessions, meetings, notes, and dictations now use consistent delete confirmations.
• Speaker labels, shared caption headers, setup sizing, and recording glow align more consistently across the app.
• Expanded localization includes right-to-left layout support for Arabic.

Requires macOS 14 Sonoma or later. Translation requires macOS 15 or later. Meeting Intelligence requires macOS 26 and Apple Intelligence availability.

September 7, 2026

NotchLive 1.10.2 improves caption reliability and recovery, adds progressive translation of timeline history, and makes Meeting Intelligence more consistent during longer conversations.

Captions and Translation

• Reduced unsupported caption endings around speech pauses and improved handling of speech boundaries.
• Choose the spoken language in apps and media to help recognition and reduce wrong-language captions, or use Automatic detection.
• Translate existing timeline captions progressively, with visible-row priority, priority for live translation, and a progress indicator.
• Original captions remain available. Changing the target language cancels pending history work for the previous target.

Recovery and Meeting Intelligence

• Improved capture recovery after sleep and display changes, and detection and repair of damaged speech-model downloads.
• The Now card keeps its last successful summary between updates and shows its update time.
• Improved context budgeting and reconciliation of saved evidence, questions, and action items.
• Reduced repeated error reporting for expected interruptions and recovery attempts.

September 2026

NotchLive 1.10.1 refines Dictation and listening feedback across the app, with safer Preferred Spellings behavior and more consistent controls when macOS uses Light Mode.

Safer Preferred Spellings

• Preferred Spellings now wait for an unprompted final pass to confirm usable speech, reducing vocabulary-driven text when Dictation is activated during silence.
• Preferred Spellings remain local vocabulary hints. Previous dictations are not used as context.
• Deterministic replacements continue to run locally after transcription and before translation.

Listening Appearance

• Wave color is now an Appearance setting shared by Dictation, captions, and Voice Notes.
• Automatic preserves NotchLive's mode-specific colors, alongside ten custom color choices.
• The floating Dictation bar stays dark and readable in Light Mode, including its progress, lock, and close controls.
• The Preferred Spellings input now has a clearer label.

August 2026

NotchLive 1.10 adds Dictation: a fast, private way to speak into focused text fields across Mac apps, translate while dictating, and recover recent dictations using NotchLive's local speech recognition.

Dictation

• Use customizable Dictate and Translate shortcuts with either hold-to-speak or press-once-to-start behavior.
• A compact floating bar supports automatic notch-aware placement or a chosen top, bottom, or corner position.
• Dictation follows the existing Microphone Language setting and selected speech model.
• Context-aware insertion handles surrounding spaces, preserves scripts that do not use spaces, and avoids adding a trailing period to standalone web and email addresses.
• Sessions can continue for up to 20 minutes, with a warning at one minute remaining and automatic insertion at the limit.

Translated Dictation

• Speak in the selected or automatically detected microphone language and insert the result in a chosen target language.
• Guided setup appears when macOS needs to download a translation language pair.
• If translation cannot complete, the original dictation remains available for recovery.
• Translated Dictation requires macOS 15 or later.

Personal Dictionary And Recovery

• Preferred Spellings provide short vocabulary hints for names, products, and phrases without using previous dictations as context.
• Deterministic When I say → Write replacements are applied locally after transcription and before translation.
• A searchable Dictations tab keeps interrupted, failed, and completed work available to inspect, copy, or delete.
• Dictation history and Personal Dictionary entries are included in NotchLive data export and import.

Caption And Model Reliability

• Active captions pause while Dictation uses the microphone and resume afterward, including cancellation and failure paths.
• Failed speech-model switches return to the previous working model.
• A genuinely stalled model switch can recover through a controlled relaunch instead of leaving NotchLive stuck.
• Pending Voice Notes can continue after a recoverable model-switch failure.

Control And Privacy

• Dictation is a NotchLive Pro feature and is off by default.
• Accessibility access is requested only after a Pro user enables Dictation or first invokes its shortcut.
• Dictation shortcuts stop immediately when Dictation is disabled or the Pro entitlement is no longer active.
• Speech recognition, dictation history, Personal Dictionary entries, and replacements remain local to the Mac. Anonymous diagnostics never include dictated text or audio.

August 2026

NotchLive 1.9.6 is a substantial reliability and multilingual update for Meeting Intelligence. Saved and live analysis now handles language changes, retries, interruptions, and partial failures without throwing away good work.

Meeting Intelligence In Your Language

• Choose Follow Captions, Original, App Language, or an available translated-caption language when reviewing a saved meeting.
• Follow Captions uses the meeting's captured caption track. App Language follows the language macOS uses for NotchLive rather than assuming English.
• Language variants reuse one grounded meeting analysis, cache independently, and stay aligned across Copy, Share, and Export.
• Chinese script variants, short CJK text, diacritics, and proper names are handled more consistently.

Safer Generation And Reanalysis

• Transient generation failures receive one bounded retry instead of repeatedly retrying unchanged input.
• If a later outline or overview phase fails, NotchLive can preserve a usable grounded note rather than losing the entire result.
• A best-effort language consistency pass repairs sections that clearly drift into another language without changing source facts.
• Incomplete localization never replaces a valid cached variant, and reanalysis updates related variants only after a complete refresh.
• Cancelled or stale work cannot overwrite the meeting or language selection currently on screen.

Live Reliability And Interface Polish

• Persistently failing live caption batches are isolated so later captions and summaries can continue.
• Pausing Meeting Intelligence preserves the draft and resumes the same meeting safely.
• Automatic Meeting Intelligence remains dormant until caption capture is ready, reducing unnecessary background work during login and service launches.
• Longer localized Focus and Timeline labels no longer overlap nearby Caption Display controls.

August 2026

NotchLive 1.9.5 is a focused reliability update for longer live-caption sessions and Timeline navigation on macOS 26.

Timeline Stability

• Fixed a high-CPU freeze that could occur when scrolling or navigating longer caption timelines on macOS 26.
• Stabilized transcript layout while preserving live following, the timeline rail, search, text selection, speaker controls, and click-to-jump navigation.
• Reduced redundant scroll synchronization work during active captioning.

August 2026

NotchLive 1.9.4 makes speech-model delivery and repair more resilient, keeps software-update reminders trustworthy, clarifies setup, and prevents abandoned MCP helpers from accumulating.

Reliable Speech-Model Delivery

• Added a verified backup source when the primary speech-model download is unavailable, times out, or leaves an incomplete model or tokenizer.
• Every backup archive is validated before installation.
• Repair preserves the previous model and restores it if replacement cannot be completed.
• Download failures now explain what happened and provide a clear retry or repair action.

Software Updates

• Updated the Sparkle update engine and hardened update-reminder state.
• Available-update reminders now remain visible across app restarts until the newer build is actually installed, then clear automatically.

Setup And Status

• Caption Display now includes the first-run drag-to-Screen-Recording guidance.
• Menu-bar captioning and recording indicators are smaller and easier to distinguish.

MCP Lifecycle

• Abandoned headless MCP helpers now exit after one hour without requests.
• Active requests are never interrupted, and every new request resets the inactivity period.

August 2026

NotchLive 1.9.3 makes the first captioning experience easier to understand and improves how NotchLive moves between its menu-bar presence and normal macOS windows.

Clearer First-Run Preparation

• The Timeline now shows the selected speech model's download and preparation state before the first captions arrive.
• The normal listening state returns automatically as soon as model preparation finishes.

Startup Preference

• Launch at Login is now off by default for new installs and remains available as an explicit setting.
• Existing users keep their saved Launch at Login preference.

macOS Window Behavior

• Improved the transition between NotchLive's menu-bar presence and its normal macOS application-switcher presence while desktop windows open and close.
• Failed activation-policy transitions now retry safely instead of leaving NotchLive in an inconsistent Command-Tab state.
• Aligned the bundled headless MCP helper metadata with NotchLive 1.9.3.

August 2026

NotchLive 1.9.2 makes long-running captions more resilient: it can release capture resources after real inactivity, rebuild audio after macOS sleep, detect unhealthy Kōkako processing, repair damaged speech models, and preserve deliberate work such as recordings, Voice Notes, and live agent follow-along.

Optional Inactivity Sleep

• Added optional 15-minute, 30-minute, and 60-minute inactivity pauses. The setting defaults to Never for new and existing users.
• Activity is measured from the selected audio source before transcription, so audible system audio, microphone speech, or either lane in Everything mode can keep captions active.
• Recordings, recording finalization, Voice Notes, manual Pause, and Stop keep their existing behavior and are never overridden by inactivity sleep.
• Added clear Sleeping, Waking, Wake, and Retry states for restarting capture deliberately.

macOS Sleep And Audio Recovery

• Active capture now suspends when macOS sleeps and rebuilds fresh system-audio and microphone pipelines after wake.
• Wake recovery is serialized and bounded to prevent duplicate capture pipelines, with actionable permission errors and limited retries for transient failures.
• Recordings interrupted by Mac sleep remain one logical session and disclose the unavailable interval in the saved session and exports.
• Record actions wait for the audio pipeline to be ready instead of beginning during recovery.

Kōkako And Speech-Model Health

• Kōkako now verifies that its system-audio reference is reaching echo processing, attempts one bounded audio rebuild after sustained failure, and uses Legacy for that session if reference health remains unreliable.
• Reference timing can adapt upward for slower Macs while the active capture session is running.
• NotchLive now verifies current speech-model downloads and identifies missing or changed files before using them.
• A new Repair action safely replaces an unhealthy model and verifies the replacement without leaving a partially replaced cache after interruption.

Meeting Intelligence, Updates, And First Run

• Improved Meeting Intelligence reliability for longer meetings, added phase-specific privacy-safe diagnostics, and explains when Apple's on-device safety protections restrict analysis.
• Manual update checks now show an explicit checking state, avoid overlapping Sparkle requests, and can resume a queued check when the updater becomes ready.
• New installs begin in Everything mode with Timeline, a 40-point floating caption size, and the top-center app switcher off. Existing preferences are unchanged.
• Translation defaults to the Mac's preferred supported language, microphone language remains Auto, and skipping microphone access leaves new users in system-audio mode.
• New stable installs launched from Applications enable Launch at Login by default; existing installs are unchanged.

Live MCP Continuity

• Active get_active_meeting polling maintains a short, content-free activity lease so an agent following a quiet live meeting does not lose captions to inactivity sleep.
• The lease expires automatically when polling stops; configured, idle, or abandoned helpers do not keep NotchLive awake.
• MCP remains local, read-only, permission-gated, and free of raw audio or partial-caption access.

August 2026

NotchLive 1.9.1 gives AI agents a stable local context layer for the work happening in meetings—including decisions made while the meeting is still live—without giving NotchLive authority to change your files or external tools.

Live Meeting Context For Agents

• Added a separately bundled, headless NotchLiveMCP helper. MCP clients no longer need to launch another copy of the NotchLive interface, create a Dock app, or open a window.
• Added Context Contract v1: structured, versioned results for saved meetings, reviewed transcripts, participant identity provenance, Meeting Analytics, Voice Notes, and the active meeting Timeline.
• Expanded the active Timeline into an incremental feed of finalized caption and participant-review changes. Clients can request only newer revisions, reconnect to the same run, and observe an explicit terminal state when the meeting ends.
• Kept compatibility with older MCP hosts through equivalent text results, legacy initialization support, and the original six read-only tool names.

More MCP Hosts, Same Contract

• Codex and Claude Desktop retain one-click setup, connection status, and safe disconnect controls.
• Existing 1.9.0 Codex and Claude entries created by NotchLive are migrated to the new helper automatically. Customized configurations are preserved for manual review.
• Added copyable standard stdio configuration for other MCP-capable hosts and agent harnesses, including local-model workflows built around Hermes, OpenClaw, Obsidian plugins, Ollama, or Qwen.
• Added a content-free diagnostic report for checking the helper path, signature, server version, transport, license availability, and access-switch state without reading meeting content.

Privacy And Authority

• Connecting a host does not grant it data access. The global AI access control and the separate active-caption control are checked again for every request.
• Turning access off denies new data requests immediately. Finalized active captions are removed from the MCP journal when access is revoked or the Timeline ends.
• Partial captions and raw audio are never exposed. The helper has no write, delete, rename, recording, microphone, model, or network tools.
• NotchLive provides meeting evidence; the connected host remains responsible for model selection, approvals, workspace changes, messages, and deployments.

Reliability And Control

• Hardened system-audio start, stop, and quit handling, including clearer ScreenCaptureKit failure classification and privacy-safe production diagnostics.
• Intentional stream stops no longer surface as capture failures.
• Added Open NotchLive on top-center hover as an opt-in setting. It is off by default, while the NotchPad/NotchLive picker continues to take precedence when enabled and available.

NotchLive MCP remains local, read-only, off by default, and included with NotchLive Pro.

August 2026

NotchLive 1.9 turns live captions into a speaker-aware meeting record you can inspect, correct, measure, export, move, and query locally.

Speaker-Aware Captions

• Added automatic remote-speaker labels on Apple silicon while microphone captions remain assigned to You.
• Added Auto, 1:1, and Group meeting context with a participant overview in the live Timeline.
• Live labels remain provisional, improve as evidence arrives, and reconcile again when retained audio finalizes.
• Speech without enough evidence remains visibly unresolved instead of being forced onto a participant.

Participant Review And Transcript Control

• Pro users can rename participants, merge duplicates, reassign or split utterances, create participants, mark speech unknown, remove irrelevant captions, and undo the latest review edit.
• Added evidence-based participant name suggestions that require confirmation before becoming authoritative.
• Added Raw and Clean transcript views so the captured record remains available while reviewed output becomes easier to read and reuse.

Meeting Analytics

• Added words, turns, valid-timing speech, speaking pace, turn duration, conversation share, participation balance, unresolved speech, and speaker-label coverage.
• Participant corrections recalculate analytics from the reviewed transcript instead of asking a generative model to estimate the conversation.
• Kept Meeting Analytics separate from Meeting Intelligence, the optional Apple Intelligence layer for focus, questions, decisions, actions, entities, and summaries.

Portable Data And AI Connections

• Added versioned data-bundle export and import for sessions, Voice Notes, participant review, and rebuildable analytics, with integrity checks and conflict preservation.
• Added an explicit, read-only local MCP server for supported Codex and Claude Desktop clients to search saved meetings, reviewed transcripts, analytics, and Voice Notes.
• Active Timeline access uses a second, separate opt-in. MCP cannot modify NotchLive data, record audio, use the microphone, access the network, or run models.

August 2026

NotchLive 1.8.5 introduces Kōkako 2.0, a rebuilt microphone path designed to reduce matching Mac playback inside NotchLive without taking over the voice processing used by meeting apps.

Kōkako 2.0

• Replaced the Kōkako 1.0 Apple Voice Processing path with normal microphone capture and local WebRTC AEC3 echo cancellation.
• Uses a local ScreenCaptureKit playback reference to help reduce matching Mac audio in microphone captions.
• Leaves Zoom, Google Meet, Microsoft Teams, and other call apps in control of the microphone processing sent to participants.
• Uses Legacy for a microphone-only caption session if the playback reference or AEC3 cannot start, without changing the saved engine. Everything mode still requires system-audio capture.

Compatibility And Boundaries

• Kōkako 2.0 performs echo cancellation only; it does not add noise suppression or automatic gain control.
• Voice Notes use the Legacy microphone path in this release and do not run through AEC3.
• The minimum deployment target remains macOS 14 on Apple silicon and Intel Macs; physical Kōkako 2.0 call and audio-route validation on macOS 14 remains pending.

August 2026

NotchLive 1.8.4 keeps captions intact around copy and export prompts, and makes background captioning, window controls, and Dock behavior easier to understand.

Live Timeline And Upgrade UX

• Copy and Export upgrade prompts no longer replace or clear captions already visible in Timeline.
• Pro prompts now offer clearer purchase, license activation, and dismiss choices.
• Fixed malformed or mirrored Copy and Export labels when Caption Display first opens.
• Kept setup guidance consistent between Timeline and Focus views.

Caption Controls And Windows

• Clarified the difference between hiding NotchLive while captions continue and stopping captions completely.
• Added clearer background-caption status and controls to the menu bar.
• Closing Caption Display now stops captioning; minimizing it or using Hide can keep captions running.
• Opening NotchLive from the Dock after a full quit now brings up Caption Display with one click without changing first-run onboarding.

August 2026

NotchLive 1.8.3 improves audio teardown, translation setup, and Caption Display reliability.

Audio And Widget

• Pausing or stopping Kōkako captioning now restores Mac audio correctly instead of leaving it reduced until NotchLive quits.
• Added an Appearance setting to hide the extra microphone-caption panel below the widget in Everything mode while microphone captions, Timeline, and saved sessions continue normally.

Translation And Voice Notes

• Fixed regional language variants such as English (UK) being incorrectly marked unavailable for same-language translation.
• Translation menus now start required language downloads, keep unsupported choices unavailable, and apply the selection after setup finishes.
• Simplified live-caption translation around “Translate to [language]” and clarified Voice Notes source-to-target controls.

Caption Display

• Timeline now shows the same interactive setup banner as Focus when caption setup is incomplete.
• Fixed clipped Copy and Export labels when Timeline opens for the first time.

July 2026

NotchLive 1.8.2 makes Live Timeline search easier to follow and gives Caption Display a more predictable keyboard, Dock, and window workflow.

Timeline Search

• Search now jumps directly to the first result and keeps the active result visibly selected.
• Matching words are highlighted clearly in both light and dark appearances.
• Move between results with Return, the search controls, or the Up and Down arrow keys.
• Press Escape once to clear the query and again to close the search field.
• Fixed later Timeline captions sometimes remaining dim or unavailable to search.

Caption Display Controls

• The Caption Display shortcut can now be customized in Settings and is included in onboarding.
• Improved shortcut activation so Caption Display comes forward reliably without reopening the notch widget behind it.
• Added Dock menu actions for Caption Display, the widget, captioning, saved sessions, and Settings.
• Made closing behavior follow how Caption Display was opened: return to the widget when opened from the widget, or stop captioning when opened from the Dock.
• Prevented resuming captions from reopening the widget while Caption Display is visible.

Mac Polish And Reliability

• Refined transitions between the notch widget and Caption Display, including behavior on macOS 15.
• Corrected oversized NotchLive Dock icon rendering on macOS 14 and 15.

July 2026

NotchLive 1.8.1 adds Live Timeline: switch from a focused current caption to searchable context without stopping the live stream.

Focus And Live Timeline

• Added Focus and Timeline modes to the floating caption window.
• Scroll through finalized captions while new captions continue arriving.
• Search the current live transcript with Command-F.
• Copy everything captured so far or export it as TXT or Markdown.
• Added source-aware Audio and You labels. These identify the capture source, not individual speakers.
• Added a micro-timeline, unseen-caption count, conditional edge fades, and Jump to Live for fast navigation.

A More Mac-Native Caption Window

• Added an optional pin that keeps captions above Zoom, Meet, and other working apps.
• Made the open caption window available through Command-Tab and the Dock, then restored accessory behavior when it closes.
• Improved compact-width controls, global Timeline text sizing, light mode, and responsive layout.

Feedback And Reliability

• Added a calmer audio-reactive visualizer for Caption, Translate, and Voice Notes modes, with Reduce Motion support.
• Improved scroll-state reliability and localized plural handling.
• Fixed duplicate live previews in Meeting Intelligence.

An unrecorded Live Timeline stays in memory for the current live run only. Record a session when you want a transcript retained after the run.

July 2026

NotchLive 1.8.0 introduces Kōkako, a new microphone engine for clearer speech capture, better audio-device recovery, and more responsive captions, translation, Voice Notes, recordings, and Meeting Intelligence.

Kōkako Voice Engine

• Added Kōkako as the recommended microphone engine, while keeping Legacy available as a compatibility option in Settings.
• Improved quiet-speech capture and added Quiet Speech, Balanced, and Noisy Room sensitivity choices.
• Added automatic per-session fallback to Legacy if Kōkako cannot start, without changing the saved preference.
• Kept speech recognition private and on-device with WhisperKit.

Reliability And Controls

• Improved recovery when connecting or disconnecting AirPods, wired microphones, and USB audio devices.
• Rebuilt the audio graph safely when switching between Kōkako and Legacy.
• Added bounded recovery when a microphone starts but produces no audio frames.
• Added an optional Hide after silence control with an adjustable delay up to 30 seconds.
• Separated system and microphone caption surfaces so one silent source no longer dims the other.
• Improved Voice Notes startup feedback, model switching, first-time model preparation, and idle transcription behavior.

Translation

• Added guided setup for missing source and target language pairs.
• Kept captions visible while NotchLive checks translation availability.
• Improved language detection, target switching, and readiness updates after installing languages.
• Preserved the original caption when translation fails instead of showing an empty result.

July 2026

NotchLive 1.7.1 improves multilingual caption reliability, script-aware text layout, the floating caption experience, and recovery when capture or Meeting Intelligence ends unexpectedly.

Multilingual Captions

• Improved native-language live captions with better language conditioning, especially when using smaller speech models.
• Added script-aware caption wrapping for Thai, Chinese, Japanese, and other languages that do not consistently use spaces.
• Improved language handling so captions stay in the spoken language before optional translation.

Floating Captions

• Redesigned the floating caption window with balanced system-audio and microphone streams.
• Added adjustable floating caption text from 16pt.
• Made Sign Cues more compact in the floating window.
• Added subtle hover feedback while preserving control tooltips.

Reliability And Diagnostics

• Improved capture rollback when audio startup fails or permissions change.
• Improved Meeting Intelligence persistence so active work is recovered and saved more reliably.
• Added optional privacy-safe diagnostics for setup and model readiness. Diagnostics never include audio, captions, transcripts, notes, or meeting content and can be disabled in Settings.

July 2026

NotchLive 1.7.0 adds Meeting Intelligence, native sharing across saved content, clearer setup recovery, searchable settings, and accessibility-focused Sign Cues.

Meeting Intelligence

• Added Meeting Intelligence for supported Macs with Apple Intelligence enabled.
• Added a live Meeting Intelligence window with current focus, key points, open questions, action items, key moments, and meeting time.
• Added saved Meetings with reanalyze, copy, share, export, and delete actions.
• Added automatic Meeting Intelligence preference controls in Settings and the menu bar.
• Meeting Intelligence stays private to the Mac experience: no meeting bot joins the call, and NotchLive does not upload audio to a cloud transcription service.

See the current AI meeting notes workflow on Mac, including live captions, Meeting Intelligence, optional transcripts, and Live Timeline.

Sharing And Saved Content

• Added native macOS sharing for Meetings, Sessions, and Notes.
• Added copied states to copy actions so users get clear feedback after copying saved content.
• Aligned Notes actions with Meetings and Sessions for a more consistent saved-content UI.

Setup And Settings

• Added Finish Setup surfaces for missing app audio permission, microphone permission, and speech model setup.
• Added drag-to-add help for macOS Screen Recording permission when the app is not already listed.
• Made the Finish Setup banner fully clickable from the NotchLive window.
• Added a searchable Settings window so users can find growing settings faster.
• Added hover tooltips for the main NotchLive toolbar controls.

Captions And Accessibility

• Improved microphone caption behavior in noisy speaker environments while preserving Voice Notes behavior.
• Added experimental Sign Cues: ASL-inspired visual cues for questions, speaker hand-offs, waiting, repeating, slowing down, action items, and name mentions. Sign Cues are not sign language translation.
• Updated the app icon assets so NotchLive appears cleanly in macOS app lists and permission surfaces.
• Changed automatic update visibility timing so update checks can surface within about 6 hours instead of waiting roughly 24 hours.

June 2026

NotchLive 1.6.3 rebuilds first-run onboarding around the NotchLive window, with clearer setup guidance for captions, translations, sessions, and voice notes.

Onboarding

• Added a new NotchLive-style onboarding flow with natural language setup steps.
• Clarified app audio, microphone access, speech model setup, and starting captions.
• Added model download progress, retry, and cancel handling during onboarding.
• Improved onboarding behavior so the NotchLive window stays out of the way until setup is finished or skipped.
• Updated localized onboarding copy across supported languages.

June 2026

NotchLive 1.6.2 includes bug fixes and improvements.

Bug Fixes And Improvements

• General reliability improvements.

June 2026

NotchLive 1.6.1 is a focused reliability update for microphone captions and recorded sessions.

Microphone Captions

• Added a Microphone Language setting, with Auto as the default, to improve live caption accuracy when you mostly speak one language.
• Moved microphone captions onto the same accuracy-first transcription path used by Voice Notes.
• Reduced early low-confidence Whisper hallucinations in live microphone captions.
• Filtered common non-speech artifacts such as silence, keyboard sounds, clicks, and other noise labels from live microphone captions.

Recording Reliability

• Improved Both mode session recording so missed microphone segments can be backfilled from the recorded mic buffer instead of being skipped.

May 2026

NotchLive 1.6.0 makes meeting context easier to capture and reuse: localized UI, Ghost Mode, meeting prompts, and export-ready TXT, Markdown, and JSON outputs.

Exports And Sessions

• Added one Export menu for saved sessions and Voice Notes with TXT, Markdown, and JSON options.
• Added a default export folder setting, with automatic exports organized into clean per-session and per-note folders.
• Added Markdown and JSON exports for notes, so dictated notes are easier to reuse outside NotchLive.
• Improved AI session transcript readability with paragraph grouping instead of one long block.
• Made older AI transcript regeneration safer for saved sessions.

Localization

• Localized the app UI across supported languages, including onboarding, Settings, release-critical labels, and Pro purchase surfaces.
• Unsupported system languages continue to fall back to English.

Privacy And Meeting Controls

• Added Ghost Mode to hide NotchLive from screen sharing and screen recordings while captions keep running.
• Added Ghost Mode controls in the notch UI, Settings, and the menu bar.
• Added ask-to-record meeting detection that can show a small branded prompt when NotchLive detects a likely meeting or call.
• Added a menu bar recording indicator so active recordings are easier to notice.

Recording Controls And Fixes

• Added the audio source switcher to the main NotchLive UI.
• Made the Notes record button match the main Caption and Translate record button.
• Fixed microphone startup when recording from Notes first.
• Fixed a settings-window interaction bug after focus changes.

May 2026

A translation reliability release that fixes macOS language-pair downloads, refreshes Settings after system language changes, and filters keyboard or fan noise before baby-notch microphone captions reach Whisper.

Translation Downloads

• Fixed translation language downloads getting stuck after macOS opened the native language-pair download prompt.
• Uses macOS English-to-target translation pair status as the source of truth instead of stale local download state.
• Added Refresh and Retry states for translation language status in Settings.
• Settings now refreshes language status after returning from macOS System Settings.

Translation Reliability

• Refreshes the active Apple Translation session after a selected target language pair finishes installing.
• Stops unavailable Apple Translation sessions from being treated as successful translations.
• Prevents untranslated source text from being cached as translated output when Apple Translation is not ready.

Caption Reliability

• Filters keyboard and fan noise before sending baby-notch mic audio to Whisper in Both mode.

May 2026

A maintenance release focused on lower idle energy use, cleaner sibling-app switching, and notch hover behavior that works consistently on notched and external displays.

Energy Use

• Replaced continuous global mouse-move polling with event-driven hover tracking around the notch area.
• Click monitors now run only while the NotchLive UI is open.
• Reduced background work when NotchLive is idle, including when captions or notes are not active.

Notch Switcher

• Expanded the shared NotchPad / NotchLive hover area so the switcher is easier to reveal.
• Matched the hover trigger to the switcher panel footprint on both notched and non-notched screens.
• Improved switcher exit tracking without returning to global mouse tracking.

Interface Polish

• Updated switcher-only app icons so the NotchPad and NotchLive selector uses clean glyph artwork without app-icon borders.

May 2026

A new compact high-accuracy speech model, smoother model switching, natural-language Voice Notes with translation controls, non-English microphone caption fixes, and Team Pack license support.

Speech Models

• Added Large Turbo Compact, a high-quality speech model option with a smaller download than the full Large Turbo model.
• Updated the full Large Turbo model entry to reflect its larger download size and highest-quality positioning.
• Improved model switching so captions can keep running while another model is prepared when a usable model is already loaded.
• Fixed stale model loading states after quick model changes, cancellations, deletes, or re-downloads.
• Improved incomplete model download recovery by clearing broken model folders and verifying required model files before loading.

Voice Notes

• Added AI, Translate, and Raw views for Voice Notes.
• Voice Notes now transcribe in the spoken language instead of being forced into English.
• Added a source-language picker for Voice Notes, with Auto plus manual language choices.
• Added a translation target picker directly in the Notes controls.
• Notes can now be translated after dictation without leaving the notch.
• Changing the Notes translation target now refreshes the visible translation instead of leaving stale text.
• Copy, save, export, and NotchPad handoff now use the selected Notes view when appropriate.
• Non-English notes now avoid English-specific filler cleanup, so Turkish and other languages are preserved more naturally.
• Fixed the Notes translation language picker becoming unclickable after note text appeared on screen.

Microphone Captions

• Fixed microphone captions in Caption mode so non-English speech is transcribed in the spoken language instead of being translated to English.
• Improved Both mode microphone captions so the baby notch uses transcription behavior and follows the selected speech model.
• Recorded microphone transcript entries in Translate mode now keep source text and translated text with the selected target language.
• Notes dictation uses the same safer microphone transcription path, including the selected Notes source language when one is chosen.

Licensing

• Added support for NotchLive Team Pack license keys.
• Team Pack 5, Team Pack 10, and Team Pack 25 keys now unlock Pro through the normal license-key activation flow.
• Existing individual NotchLive Pro keys continue to activate and revalidate through the original Pro path.
• Valid Team Pack keys continue to revalidate as NotchLive Pro licenses.
• NotchPad and other wrong-product license keys still reject instead of unlocking NotchLive.
• Settings now shows the active plan label for individual and team-pack licenses.
• Improved activation reliability for valid NotchLive keys when the purchase service omits plan details in a successful response.

Settings And Window Fixes

• Fixed a Notes-path Settings bug where the Settings window could become hard to close or interact with.
• Settings now stays clickable when opened from Notes.
• Fixed Settings model rows getting stuck in stale download, selected, loading, or ready states after model changes.

May 2026

A smoother first-run experience, stronger model reliability, automatic audio-device recovery, and a more polished macOS interface.

Smoother Setup

• Redesigned onboarding so new users understand permissions, choose a speech model, watch it download and load, and start captions from a clearer final screen.
• Onboarding now resumes after macOS asks to restart NotchLive for Screen Recording permission.
• The first caption run now uses the model prepared during onboarding instead of getting stuck in a loading state.
• Added a shortcut summary for Show/Hide, Start/Stop, mode switching, and Copy Last.

Reliability

• Fixed cases where the notch could stay on Loading model... even though the model was already loaded.
• Model downloads and loading now detect incomplete files, check tokenizer availability, and time out with clearer errors instead of hanging.
• Connecting AirPods or changing the default microphone/output device now restarts capture automatically without relaunching or toggling sources.
• Settings model switches now reuse an already-loaded model when possible.

Interface Polish

• Refined settings, sessions, notes, and detail windows with a more native macOS 26-style glass treatment where it helps.
• Improved segmented controls, back buttons, export/delete actions, and other controls so the full button area is clickable.
• Cleaned up onboarding animations, permission cards, model rows, shortcut chips, and final setup messaging.
• Updated the app icon and release DMG icon so NotchLive looks right on pre-Tahoe macOS as well as newer macOS versions.

May 2026

A small bug fix release for the shared NotchPad / NotchLive selector and gentler scheduled update reminders.

Fixes

• The shared NotchPad / NotchLive selector now stays open while moving from the notch hover zone into the selector on notched Macs.
• Scheduled update checks now show the notch update button instead of interrupting with an automatic dialog in the background.
• Clicking the update button still opens the normal Sparkle update flow and release notes.

April 2026

On-demand AI dictation for saved sessions, better transcript exports, and clearer update reminders.

Features

• Saved sessions now have a Raw / AI switch so you can keep the original transcript or generate a cleaned dictation view when you need it.
• AI dictation is generated only on demand, runs through the existing local cleanup flow, and is saved with the session after generation.
• Session AI keeps speaker turns cleaner: microphone entries become You:, captured audio becomes Audio:, and system-only sessions stay plain.
• The notch now shows a visible update reminder when Sparkle finds an update, so updates are not buried in Settings.

Export

• Raw TXT export keeps timestamps and YOU labels.
• AI TXT export saves the generated AI dictation with an AI filename suffix.
• JSON export includes saved AI dictation data when it exists.

April 2026

Recording reliability, Both mode mic caption stability, and cleaner note feedback.

Fixes

• Recording stop, pause, and quit now finalize more safely so final session audio and ended timestamps are preserved.
• Saved sessions now persist translated captions more reliably.
• Stale caption lines clear more predictably after silence and when switching between widget and screen caption modes.
• Both mode microphone captions now show a loading state while the mic model is preparing and recover more reliably after audio source changes.
• Notes copy and save actions now show visible confirmation, and note cleanup preserves meaning and line breaks better.

April 2026

Voice Notes for Pro — dictate, clean up, save, export, and send notes to NotchPad.

Features

• Voice Notes — a focused dark notepad for Pro users that turns microphone input into speech-to-text notes.
• AI / Raw views — review the natural transcript or switch to an AI-cleaned note that reads more like something you wrote.
• Saved Notes library — keep dictated notes in NotchLive, copy them, export them, or delete them later.
• Send to NotchPad — hand finished notes to NotchPad when you want them in your permanent notebook.
• Sibling-app handoff polish — NotchLive now backs away before opening NotchPad so the two notch panels do not stack on top of each other.

April 2026

Baby NotchLive — compact mic captions panel in Both mode.

Features

• Baby NotchLive — when audio is set to "Both", a compact secondary panel now appears below the notch showing your microphone captions in real time. No more opening the companion window just to see what you're saying.

April 2026

Microphone permission fix.

Fixes

• Fixed microphone permission not working — the app now correctly requests microphone access

April 2026

Bug fixes and improvements.

Fixes

• Improved reliability around activation and purchase flows.
• General stability improvements.

March 2026

Bug fixes and improvements.

Fixes

• Fixed Pro mode activation edge cases.
• Resolved intermittent crashes during extended captioning sessions
• General stability and performance improvements

March 2026

5 AI models, 20 translation languages, pause/resume, and redesigned Settings.

Features

• Pause and resume captioning from the widget control bar
• 5 AI models — added Medium and Large Turbo for best accuracy
• 20 translation languages — added Dutch, Hindi, Indonesian, Polish, Russian, Thai, Ukrainian, Vietnamese, Chinese (Traditional)
• New 2-column language picker with download status indicators
• Redesigned Settings with friendlier labels and model cards
• Hot-swap AI models while captioning is running
• Small model is now the default for better out-of-box accuracy

Fixes

• Better mic accuracy in Both mode
• Fixed model loading and download UI glitches

March 2026

Bug fixes and Pro upsell improvements.

Fixes

• Improved activation reliability.
• Fixed deactivation reliability.
• Fixed Pro upsell banner not showing after license changes

March 2026

Pro upsell, non-blocking translation, and bug fixes.

Features

• Pro upsell banner with Upgrade and Activate Key buttons
• Auto-dismiss Pro upsell when license is activated
• Non-blocking translation — captions appear instantly

Fixes

• Session recording no longer cuts off when changing translation language

March 2026

Caption Display improvements.

Features

• Caption Display now auto-starts captioning when opened — no extra steps needed
• New window button in the notch control bar for quick access to Caption Display
• Dual-stream captions in Both mode: your mic (cyan, top) + system audio (white, bottom)

March 2026

Caption Display window for screen sharing.

Features

• Caption Display — floating, resizable window with large captions for screen sharing in Zoom, Meet, StreamYard, and more
• Toggle with ⌥ ⌘ F or from the menu bar
• Auto-hides notch widget when Caption Display opens, restores on close

February 2026

Both mode overhaul, mic improvements, and post-session transcription.

Improvements

• Fixed "Both" mode - system audio captioned cleanly, no mic interference
• Mic gain boost (3x) for normal speaking distance
• Post-session mic transcription with "YOU" tags in session detail
• Chronological interleaving of entries
• Widget drop shadow
• Processing indicator during mic transcription

February 2026

Caption quality and UI improvements.

Improvements

• Fuller caption lines - text fills the full width before breaking to a new line
• Reduced hallucinations with improved noise filtering
• Better text spacing and readability near the notch area
• Added Website & Support links in Settings

Fixes

• Fixed auto-update feed so future updates are detected automatically
• Fixed microphone not working in release builds

February 2026

Live AI captions and translated speech in your MacBook's notch.

Features

Captions

• Real-time speech-to-text powered by on-device Whisper AI
• Three model tiers: Light (~40 MB), Standard (~140 MB), Pro (~460 MB)
• Automatic language detection for 90+ languages
• Streaming transcription with partial updates - text appears within ~1.5s of speech

Translation

• Translated speech in 11 languages via Apple Translation (macOS 15+)
• Debounced partial translation - see translations as you listen, not just after silence
• Supported targets: English, Spanish, French, German, Chinese, Japanese, Korean, Portuguese, Italian, Turkish, Arabic

Audio Sources

• System audio capture via ScreenCaptureKit - works with any app (Zoom, Meet, Teams, YouTube, etc.)
• Microphone input for in-person conversations
• Switchable on the fly

Notch Widget

• Floats over the notch area with smooth expand/collapse animations
• Works on MacBooks with and without a notch
• Hover to reveal controls: mode switch, language picker, recording, settings
• Auto-dims after silence, instantly un-dims when speech resumes

Session Recording (Pro)

• Record transcription sessions with timestamps
• Searchable session history with rename, export, and delete
• Export as TXT or JSON
• Breathing red edge overlay indicates active recording

Privacy

• Fully on-device - audio never leaves your Mac
• No cloud audio processing, no required account, and no uploaded audio or transcripts
• Hardened runtime, code signed, and notarized by Apple

Keyboard Shortcuts

• ⌥ ⌘ C Toggle captioning
• ⌥ ⌘ T Cycle mode (Caption / Translate)
• ⌥ ⌘ K Copy last caption to clipboard
• ⌥ ⌘ H Toggle widget visibility

Other

• Menu bar app (no Dock icon)
• Guided onboarding for permissions and model download
• Automatic updates via Sparkle
• Free forever for captions. Pro $14.99 one-time for translation, sessions, and export.