News and usage data in your language
News, the usage data consent prompt, privacy explanations and transcription modes are now translated into all 27 interface languages. The app’s release history and website use the same translations.
Read moreMemoxy
Memoxy news
Product updates, release notes and ways to share your feedback.
News in your profile
Your profile now includes news: changes in your installed version, product updates and invitations to share feedback. Read full articles inside Memoxy, share their public links or open them in your browser. Read and dismissed markers stay on this device. Delivery of new website posts requires deployment of the updat…
Read moreUsage data choice on first launch
Memoxy asks you to allow or decline usage sharing on first opening. Your choice survives app updates; Later postpones the question until the next launch. You can change it in Settings. Changes to data categories or purposes require new consent.
Read moreDevice IDs
Memoxy saves a device ID on first launch, before sign-in. You can find the same ID in the app profile and account dashboard.
Read moreOptional usage data: pilot
Settings now let you separately opt in to daily usage summaries to help improve Memoxy. Sharing is off by default; data categories and retention appear before your choice. Turning sharing off stops collection and requests deletion for this installation, including after reconnecting. Summaries exclude your content an…
Read moreNews, close to your workspace
Read product news and release notes from your profile. Longer articles open inside Memoxy, with a link to the same story on our website.
Read moreDevice IDs
Memoxy saves a device ID on first launch, before sign-in. You can find the same ID in the app profile and account dashboard.
Read moreDiagnostic export
Diagnostic reports use metadata from the current account. On Mac, JSON is saved with the system dialog; cancelling does not show a success message.
Read moreBackground account checks
Your saved profile stays available while the account is checked in the background. Waiting for another account operation no longer shows a warning.
Read morePronunciation using the letters of the translation language
AI generation of language cards now offers three transcription modes: None, Standard (IPA), and In the translation language. The last option adds an approximate pronunciation of the source text using the letters of the selected translation language.
Read moreImage processing dependency update
Updated the sharp image-processing dependency for the app and website to a patched version.
Read moreBuild dependency update
Updated the source-map-js dependency for the app and website to a patched version. Dependency security checks pass again.
Read moreSystem speech preparation counter
The system speech preparation panel on iPhone receives the number of completed recordings when preparation resumes and the current counter when task launch is delayed. If iOS rejects the background task, its reason is recorded in diagnostics; the panel still depends on the system accepting the task.
Read moreAI revisions with set context
New AI sets retain the original task and up to the two most recent accepted instructions, limited to 1,200 characters per request. These are included in later editing and generation requests, even after reopening the set. The current instruction takes precedence. Older sets without a saved task have no original prom…
Read moreRestart the set from the card corner
The set restart button is now a single icon in the top-right corner of the card. The circular arrow starts the set again, while the skip-back icon repeats the current card. The separate button row and the duplicate button shown after playback finishes have been removed.
Read moreNo duplicate task panel
On iPhone, the duplicate top panel showing background tasks, generation results and speech progress has been removed. Routine notices about applying a set or restoring a draft no longer appear at the top of the screen either. Speech preparation still appears in the system indicator.
Read moreRequesting low reasoning effort
Creating and editing sets through OpenRouter now requests low reasoning effort. The app does not increase it automatically. The model or provider may ignore unsupported parameters; actual usage depends on the selected model and provider.
Read moreMore responsive swipes
Swipes in cards, the wizard and mobile menus now respond to shorter movements.
Read moreSpeech and controls on iPhone
Preparing a set with audio shows the cause of errors and reuses ready, valid clips on retry while the dialog remains open. Newly generated WAV speech is encoded locally as compact AAC on iPhone.
Read moreSpeech preparation fixes
Fixed the request format for Gemini 3.8 TTS: instructions are sent separately from card text. Previously saved incorrect audio needs to be prepared again.
Read morePlayback without extra notifications
Playback starts without routine notices. Use Start again at any time, or enable Repeat set for continuous playback without an end message.
Read moreQuestion hints
In text checks, you can listen to the question. In listening checks, you can reveal its text as a hint. Text hints are hidden again when you change questions.
Read moreSet groups and batch actions
Select multiple sets in a wizard to share or delete them, create groups, and add or remove group membership. A set can belong to several groups. Deleting a group keeps its sets.
Read moreOrder and completion notices
Restore the original card order after shuffling. iOS shows the playback position and can notify you when speech preparation, generation or playback finishes, if notifications are allowed.
Read moreSpeech preparation and background tasks
Added progress, pause, resume and cancellation controls for speech preparation. On iOS, background work uses system authorisation; if unavailable, the app reports the limitation.
Read moreSpeech preparation and card deletion
If Gemini’s content check rejects a speech fragment, the request is retried once with an explicit reading instruction, preserving the text and voice. Preparation continues with the remaining fragments, prevents automatic screen locking, and resumes an interrupted network request when you return to the app. Card dele…
Read moreSpeech request language and rejection reasons
Gemini speech requests now include the card language and separate the transcript from reading instructions. Content moderation refusals show the provider’s reason instead of suggesting incorrect model or voice settings.
Read morePreparation and playback error details
Speech preparation errors show the cause, stage and failure code. iPhone diagnostics retain error codes without keys or card text.
Read moreSpeech preparation in Source
Speech preparation for the working set is back in Source, next to Share. Preparation can be cancelled and retried after an error.
Read moreBackground playback on iPhone
On iPhone, OpenRouter audio and saved set recordings are transferred to the native player before playback, including pauses and repeats. Extended playback with the screen locked is still being tested.
Read moreReliability and security fixes
Licences now respect their expiry date, removed OpenRouter connections delete stored keys, and recovery is safer after playback, AI generation, export or storage errors.
Read moreCheck knowledge across sets
Check the working set, selected sets or random cards from your library. Practise all mistakes or those from selected sets; listening questions use their source set’s language.
Read moreA clearer knowledge check start screen
The set, new test settings, start options and history are grouped clearly. Start buttons are equally wide and include icons; an unfinished attempt appears separately.
Read moreMore accurate text clipping warnings
Before video export, cards are checked with the fonts used in the final image. Warnings remain in the export status; they are not shown for audio exports.
Read moreSimpler knowledge checks
Start or resume a check from one screen. Repeat mistakes directly from the result; detailed answers and history are available separately.
Read moreRandom tests based on your progress
Start a random test from the current set, with more emphasis on untested cards, mistakes and cards not practised for a while. Each attempt is saved separately.
Read moreProgress and mistake practice
View first and final results by test method and practise mistakes from the current cards in a new numbered attempt.
Read moreMore reliable test history
Cards saved through the editor keep their identity across subsequent edits. Database upgrades and history backup restoration preserve existing attempts and answers.
Read moreListening tests and answer history
Choose a listening or reading test. Separate numbered attempts keep your answers and corrections, and unfinished tests can be resumed.
Read moreImport .memoxy files on iPhone
Fixed selection of .memoxy files in the iPhone import picker. Sets containing text and narration can be imported through the wizard.
Read moreA separate step for adding material
In the edit wizard, adding material is a separate step after the name and languages. You can skip it or save your changes and close the wizard. The flow follows set creation; Back and swipes preserve the draft.
Read moreClearer tests and sharing sets with audio
The wizard opens the playlist with editing off. Level, content type and card count are visible immediately; provider and model are in additional settings.
Read moreMore responsive swipes
Swipes in cards, the wizard and mobile menus now respond to shorter movements.
Read moreShare a set
Source now has a Share button to share the working set as a plain text file.
Read moreQuick card editing
The playlist editing toggle is back in Source. On the first set-editing screen, the action to continue to the cards stays visible alongside Save and close.
Read moreCreating and editing sets
Source now contains the working set selector, New, Edit and Delete actions, and the playlist. The shared wizard lets you change the name and language settings, edit cards, and add or replace material.
Read moreWarnings after editing cards
Long text on two-sided cards uses the available space on the front. Warnings are recalculated, and you can filter the playlist to review the remaining warnings.
Read moreClear validation and bulk card actions
The playlist shows the count and numbers of incomplete cards, marks empty fields and lets you show only the affected cards or exclude them with undo. The reason for blocking is shown beside the save button. Import preserves row order. Text overflow warnings do not block saving: skip them all or use two sides for aff…
Read moreConsistent steps for preparing material
The wizard starts with choosing an action and reviewing settings. Pasting, importing and generating lead to the shared playlist editor, then to creating, appending to or replacing a set. Going back preserves the prompt, cards and save options. After saving, you can make the set your working set or keep the current s…
Read moreStep-by-step set creation
A new wizard guides you through choosing a set type, pasting, importing or generating material, reviewing the playlist and saving. After saving, you can review the cards or listen to the set.
Read moreBackground music during card playback on iPhone
During autoplay with the system voice, background music plays alongside speech and follows card playback when paused, resumed or stopped. The settings demo stops before cards start.
Read moreSeparate settings for language and general modes
Each mode remembers its own settings. General mode uses one language for both sides of a card; the language pair stays in the language profile. On first launch, the general-card language and translation language match the interface language.
Read moreManual language pair selection
Choose languages in Source or settings. The pair is saved with the set and restored when you open it. Text import, playback and export no longer detect the language automatically.
Read moreClearer Free limits
Free shows a limit of 100 cards per set. Paste, import and adding cards check the limit before changing data; existing larger sets are preserved.
Read moreClear clipboard saving options
When pasting, choose a new set, add to the working set or replace it. New sets get a suggested name; replacement requires confirmation and lets you change the name. The result is saved immediately.
Read moreRevisit test questions
During a knowledge check, return to earlier questions with a button or swipe and correct an answer. “Next” saves the change, and the result reflects corrected answers.
Read moreWorking set and AI draft
The switch is now a pair of tabs, “Working set” and “AI draft”, directly above the cards. The active tab is underlined; switching shows the corresponding list without replacing the working set.
Read moreSaving and deleting sets
Saving changes and deleting a set now use confirmation dialogs inside the app. Once saved, the new name appears in set lists and during playback; unsaved name edits no longer change the playback title.
Read moreEasier answer selection and reading
Options show their number and text. Tap to open the full card, select a radio button, then press “Next” to record your answer. You can change your choice before confirming.
Read moreKnowledge check pilot
Test your working set with shuffled questions and up to four answers. Expand long answers separately from choosing them; see your score and mistakes at the end.
Read moreMore room for cards
On iPhone, the card count and estimated set duration share a compact indicator at the top right. The permanent information block below the playback controls has been removed.
Read moreSource gestures and a clearer profile
On iPhone, swipe right in Source or down on its header to return to Play. Fields and playlist scrolling retain their own controls; the working set stays intact.
Read moreGestures and settings headers
Improved recognition of short swipes and imprecise starts in mobile menus and cards. Menu navigation rows support swiping without also activating the row.
Read moreCompact settings fields
Listen buttons now sit beside music, volume, language and voice fields, with labels above. Full and Quick settings share this layout wherever listening is available.
Read moreControls matched to their purpose
Checkboxes beside each settings group now control whether it appears in Quick settings. They change only the contents of Quick settings, not the features themselves.
Read moreMobile gestures
In Settings, swipe right to go back and left to reopen the last section. At the menu root, swipe right to close; swiping down on the header also closes Settings, Profile and Quick settings.
Read moreClearer card side controls
Choose one or two sides below Working set. Each set saves its own mode; Custom automatically indicates cards whose layout differs.
Read moreIndividual card sides
Choose one or two sides for an individual card, or keep the set's setting. The choice is retained when editing, saving, reopening and exporting the set.
Read moreCard clipping warnings
Text still clipped after automatic sizing is flagged above preview and in the playlist. You can keep it as is, fix it or delete the card with an option to undo.
Read moreLong-word wrapping
Automatic sizing first tries to keep each word whole by reducing the font size within the allowed limits. This also applies to previously saved cards.
Read moreAutomatic text fitting
When text does not fit, the app tries smaller sizes within the allowed limits before truncating it, consistently in preview and export.
Read moreAccurate text truncation detection
An ellipsis in the original card text no longer triggers a false truncation warning.
Read moreWorking set audio and shared build numbering
Offline audio preparation and its progress now appear in Source, below the working set selector.
Read moreClearer settings
Voice settings are grouped with one Quick Settings shortcut; diagnostics are collapsed and redundant actions removed.
Read moreStartup when Keychain is unavailable
A Keychain error no longer blocks startup on Mac. An unfinished migration can be retried.
Read moreClearer forms and simpler sections
Removed nested decorative panels from settings and AI results. Result actions appear before the collapsible editor, and long lists scroll with the page.
Read moreA simpler AI workspace and restored test access
In test builds, saved Advanced access is used for AI requests again after an update. Production licence verification is unchanged.
Read moreAI on iPhone and organised sets
iPhone now supports OpenRouter AI generation and suggested card edits. Internet, a configured key and licence access to the feature are required; the model can be entered manually.
Read moreSystem voice when TTS is unavailable
If external TTS is unavailable and the audio is not already cached, playback continues with the system voice.
Read moreHeading import and stable touch fields
Files with # headings can be split into named sets or kept as one set with title cards. Imported heading text stays unchanged and uses the app’s styling.
Read moreSimpler speech provider selection
Active TTS now offers only System Voice and OpenRouter on every platform. OpenRouter defaults to Gemini Flash TTS and still accepts manually entered model IDs.
Read moreiPhone connections and shared alignment
The iPhone connection editor now supports OpenRouter, Google TTS and ElevenLabs. Successful speech responses are stored on device for reuse; external-account playback and offline operation still require device verification.
Read moreReliable iPhone playback and unified controls
The iPhone playback queue now restarts reliably after Stop and when moving between regular and fullscreen views. Now Playing keeps card artwork, metadata, and system controls in Control Center and on the Lock Screen.
Read moreNative iPhone playback and clearer cards
The unreleased iPhone app now uses the installed system voices through native speech, applies the selected voice and pace, keeps the display awake during active playback, stores secrets in iOS Keychain, and bundles temporary background-music tracks for offline device testing.
Read moreFullscreen playback controls
The unreleased iPhone fullscreen preview now keeps the same five-action playback bar at the bottom; the fullscreen toggle remains last, and the bar stays above the system safe area.
Read moreAdaptive iPhone preview
The iPhone preview now expands to the available space while preserving its aspect ratio and margins; the fullscreen action is the final control in the playback bar.
Read moreMore room for iPhone study cards
The unreleased iPhone workspace now places Settings and Profile beside Play and Source in one bottom area, removes the top bar, and keeps content below the system safe area.
Read moreMore reliable desktop startup
The packaged desktop app now loads its built-in interface independently from the local service and shows a recoverable startup error instead of an empty window.
Read moreClear speech speed controls
Word and translation speed now use four distinct presets: Very slow, Slow, Normal, and Fast.
Read moreBounded scrolling for generated results
Large generated results now keep card and proposed-change lists inside their own scroll areas, so hidden rows no longer create an empty tail on the main workspace.
Read moreWorkspace layout and card text scale
Source and AI panels now keep their own content height in wide layouts, removing the large empty area below the last AI activity.
Read moreFocused language and provider settings
Used languages now combines the interface-language choice with a persistent working-language list; the current project pair remains available, while language selectors and voice mappings hide unused languages.
Read moreGeneration performance metrics
Recent activity now shows a live elapsed timer during generation, then records the date and time, model, generated word or phrase count, total duration, and average time per card in one compact row.
Read moreCompact profile, settings, and card editing
Profile now uses a compact content-height layout with a clearer hierarchy for identity and license details.
Read moreCompact generation structure settings
Transcription, Level, Type, and Count now appear in that order on one row when the window is wide enough.
Read moreTest Advanced activation in internal builds
Beta/open builds now provide a dedicated action to activate test Advanced without a signed license.
Read moreReviewed AI editing for generated cards
Generated results now use a structured card editor with selection and mass deletion for several cards.
Read moreClearer AI generation workflow
The AI workspace now explains which current or saved cards are used to prevent duplicates and shows the unique card count.
Read moreAI providers and Free access
AI Features now uses only OpenRouter or Ollama; the Local Template AI option has been removed.
Read moreBuild number in the profile
The profile now shows the installed Memoxy version and build number, while license actions remain accessible in compact windows.
Read moreClear AI provider selection
The active AI generator can now be selected directly beside the generation settings.
Read moreWatermark removed after Advanced activation
The automatic Memoxy watermark now disappears after Advanced activation; a custom watermark can still be enabled in video settings.
Read moreAI generation for Advanced
AI generation is now available in production builds after Advanced activation; OpenRouter or Ollama still needs to be configured separately.
Read moreAdvanced activation and first launch
Advanced activation codes are now accepted correctly in production builds.
Read moreMemoxy
The app, website, exported files and Free watermark now use the product name Memoxy.
Read moreClearer Memoxy website
The Russian website now explains video and audio cards, AI and voice setup, privacy, Free versus Advanced, and beta access in a clearer task-led order.
Read moreClear Free and Advanced watermarks
Free preview and export now always keep a subtle Memoxy pattern across the card; Advanced can customize or disable it after activation.
Read moreClear API key saving
OpenRouter now has a separate Save API key button beside the key field, while model and voice settings remain in the profile save action.
Read moreConsistent Memoxy desktop package
The macOS app, DMG, window, menu, exports, watermark, and OpenRouter integration now use the single product name Memoxy.
Read moreAdvanced card limit and temporary shuffle
Free exports remain limited to 100 cards, while Advanced can export up to 1,000 cards in one video or audio job.
Read moreTarget-language and voice-provider matrix
The internal language registry now covers 61 target language and script variants across Latin, Cyrillic, Japanese, Korean, Simplified Chinese, and Traditional Chinese.
Read moreSystem language and multilingual groundwork
On a new installation, the app now uses the first supported macOS language and falls back to English; an explicitly selected language is still preserved.
Read moreCard limits for Free and Advanced
Free can now export up to 100 cards from one set, while Advanced removes the plan card limit for video and audio exports.
Read moreClearer free export watermark
Free exports now use a repeated diagonal Memoxy watermark in preview and rendered videos.
Read morePronunciation comparison presets
Speech speed selection now keeps only the useful presets: Very slow, Slow, Normal, and Fast.
Read moreAI generation diagnostics and draft validation
AI prompt history stays above Generate draft, while LLM status, results, and errors sit below it with provider/model diagnostics.
Read moreAI generation, OpenRouter, and desktop build fixes
AI generation now keeps a local history: previous results can be opened from History AI, reused, or cleared.
Read moreBuild journal
Added an in-app change log in Settings.
Read more