Release notes

Changelog

Everything new in DijiFlow Dictate, version by version.

macOS

macOS 14+ · Apple Silicon · 8 GB memory

v8.1

Latest2026

A simpler app, with one card for each language you dictate in.

  • New Languages tab: each language has its own card with its shortcuts, and Pro and the trial can add up to four more.
  • A shorter first setup: choose your languages, download once, then try a sentence.
  • Simple chooses the model for you, and each download shows its size before it starts.
  • Fixed cases where part of a long dictation could be missing in languages such as Hindi and Thai.
  • Long dictations with "Fast" are more accurate.
  • When your Mac runs critically low on memory, the app frees memory by itself.

The Languages tab

  • The first tab is now called Languages. A new setup shows one card, titled by its language, with the spoken language, the output language and both shortcuts.
  • Press "Add a language" for another card. Pro and the trial have up to five, the free version has one.
  • Clearing a shortcut turns it off, and a shortcut that is off leaves its keys free for another one.
  • Notices and the menu-bar menu name each card by its language.

Simple and Advanced

  • Simple chooses the model for each language and downloads what a choice needs with one press, with the size shown first.
  • Advanced offers "Large", "Express" and, for the languages it covers, "Fast" on each card, and more settings.
  • Fine-tuning controls that the app now sets by itself have left the Settings window, and their values return to the defaults once.
  • Hold Option while you open the menu-bar menu to reach Expert settings, which has the Models tab.

A shorter first setup

  • The first screen asks for the language you speak and the language you want typed, and shows the download size for both.
  • The next screen lets you hold the shortcut, say a sentence and see your words.
  • The app asks for the microphone when you choose to try it, and offers the Accessibility permission after the try, with its reason.
  • A switch on the first screen lets DijiFlow start when you log in, so your shortcut always works.

Downloads and memory

  • Interrupted downloads resume where they stopped.
  • Before a download starts, the app checks that there is enough free disk space.
  • When your Mac runs critically low on memory, the app frees memory by itself and can use "Express" in place of "Large". Your chosen model stays saved.

Vocabulary and files

  • On the Vocabulary tab, Simple shows one switch per topic for all your languages, and a language you add gets the topics that are on for the others.
  • On the Transcribe File tab, Simple chooses the model for you and shows one download line.
  • The file tab shows whether a run uses the Neural Engine or the processor.

Reliability

  • Fixed cases where part of a long dictation could be missing in languages such as Hindi and Thai.
  • Fixed rare cases where a closing phrase that was never spoken could appear at the end of a Japanese dictation.
  • A model change that is cancelled or cannot finish, for example after a failed download, keeps the recording in progress.
  • Fixed rare cases where a held shortcut could end a recording early after a settings change.
  • Fixed a rare crash on Macs with no microphone, and a rare crash during the first setup.

Sounds, accessibility and privacy

  • The switch for recording sounds is in Advanced.
  • Pre-press capture has its own switch in Advanced. It is off after the update, even if you had it on; turn it on again there if you use it.
  • Every switch in Settings has a label for VoiceOver.
  • Usage statistics also count, once each, three steps of the first setup: the test dictation, the first text delivered to another app and the Accessibility permission.

v8.0

Major update2026

New models, subtitles in the Free version and more reliable dictation.

  • New "Express" and "Fast" models in Pro and the trial, rated for accuracy, speed and memory.
  • "Fast" covers 16 languages, "Transcribe live" included, and file transcripts in 15.
  • Fixed rare cases where "Fast" could switch to another language mid-dictation.
  • Better speaker identification in files and subtitles.
  • Fewer words that were never spoken during silence or music.
  • One- or two-word dictations come without a closing period or capital.
  • More reliable vocabulary correction in all 29 languages.
  • Faster first dictation after launch, with a "Loading model..." notice.
  • Fixed rare cases of dropped quiet speech, brief pauses and delayed pastes.
  • New: anonymous usage statistics, on by default; switch them off at the bottom of Settings.

A new first screen

  • On a new Mac, the first screen now highlights the 30-day free trial of every Pro feature, with no card and no email.

Subtitles in the Free version

  • The Free version makes subtitles for the first 3 minutes of each file, as it does for file transcripts.
  • Translated subtitles are part of Pro.

Model changes

  • When "Fast" is unsure of a few sentences and "Express" is not installed, a notice suggests downloading "Express" to check them.
  • Pro and the trial now offer three models: "Large", "Express" and "Fast". Older models are retired, and copies already installed keep working.
  • The Free version lists the Pro models as locked options.
  • When you switch models, the previous model stays in use until the new one is downloaded and loaded.

Subtitles that read naturally

  • Subtitles split at natural points for each language, so words that belong together stay together.

When the trial ends

  • Before the trial ends, the model for the Free version downloads in advance, so the switch to the Free version needs no internet connection.
  • A short notice explains that the trial has ended.

Performance and protection

  • Long dictation sessions run more steadily and use less memory.
  • Stronger protection for the app.

v7.19.3

2026

More complete transcripts and subtitles, and cleaner subtitle lines.

  • Fixed rare cases where a passage of a file transcript could be dropped in quiet audio or around music.
  • Long dictations and long files are checked for stretches that came back too short, and completed.
  • Fixed a rare case where subtitles could drop words between two passages.
  • Improved subtitle line breaks, so a single word is not left on a line of its own.
  • Auto-detect identifies the spoken language of each recording.
  • New: transcribe or subtitle many files in one go. Pick several files and the app works through them one by one, showing which file it is on. Transcripts are joined into one text, subtitles are saved per file.

Transcripts and dictation

  • Fixed a rare case where a passage of a file transcript could be dropped when the audio was quiet.
  • Fixed a rare case where music, applause, or singing could remove the speech around it from a transcript.
  • In long dictations and long files, a stretch that came back too short is checked again and completed.
  • With automatic language detection, a completed passage is written in the language of the recording.
  • With Auto-detect as the spoken language, a recording is transcribed in the language that was spoken.
  • Files added together now run in name order, and you can sort them or drag them into the order you want.
  • Several files can be combined into one recording with one set of speakers, or saved as separate files to a folder you choose.

Subtitles

  • Fixed a rare case where subtitles could drop words between two passages.
  • Improved line breaks, so a single word is not left on a line of its own.
  • An opening word such as "Now," is kept with its sentence.
  • In English subtitles, word pairs such as "so that" and "and then" are kept together.
  • Subtitle translation into Japanese, Korean, Ukrainian, Finnish and Hungarian keeps each line with its own sentence.
  • Fixed rare cases where a translated subtitle line could stay untranslated or come back in the wrong language.
  • Translated subtitles no longer start with stray list numbers or letters, and a sentence that repeats in a file is translated the same way each time.

Fixed

  • Fixed a rare case where the app could stop responding for a moment when the file tab was opened.
  • The "Select a File" button shows that it is opening while the file window is prepared.
  • Fixed the dictation overlay flickering when several takes overlapped.
  • Fixed a rare case where a few words could appear twice in a long recording.
  • Improved the readability of the buttons on the finished-file notice in dark mode.
  • Fixed a case where the speaker identification and translation models showed as needing a download again after an update.
  • When a model is needed, the file tab and Subtitles mode now offer the download in place, and the models list updates by itself after a download.
  • The app stays smooth while a model downloads.

Dictation

  • Dictated text is delivered reliably even when the key for the next slot is already held down.
  • When the output language cannot be changed, the reason is shown under the picker.

Interface

  • You can reopen the main window at any time from the Window menu or with ⌘0.

v7.19.2

2026

Reliable microphone selection, faster launch, and clearer memory handling.

  • Fixed a case where choosing the built-in microphone while another one was connected could end dictation with a recording error.
  • iPhone and Bluetooth microphones get the time they need to start.
  • With the lid closed, the app points out that the built-in microphone is off.
  • The app is responsive right after launch, and the file window opens without delay.
  • When a model does not fit in memory, the app explains why and offers Dynamic Memory Optimization or a smaller model.
  • Fixed rare cases where the last sentence, the closing sentences of a long dictation, a sentence after a pause, or a short dictation with digits could be dropped.
  • Fixed rare cases where a microphone notice appeared without cause or a microphone stopped responding until the app was reopened.

v7.19.1

2026

Maintenance update for 7.19: clearer progress, simpler names, and subtitles that are easier to read.

  • Burned-in subtitles are easier to read at any video size, including small clips.
  • File transcription shows a progress bar for each step, with a time estimate based on measured speed.
  • Choosing an output language checks that its translator or language pack is downloaded, and offers it before the run starts.
  • Speaker identification is available for subtitles, and the model can be downloaded from the file tab.
  • Models and translation engines are described by what they do.
  • Improved contrast in dark mode.

Files and subtitles

  • The run button names what it creates, for example an MP4 with burned-in subtitles or an SRT file.
  • Switching between Transcript and Subtitles keeps your model, or moves to the largest you have.
  • Subtitles mode is visible in Free, with its Pro features marked.
  • A finished run leaves its result card in place, with any problem stated on it.

Downloads and models

  • Model cards show what is downloaded; several can download at once and the pickers stay usable.
  • Download prompts show the size, and progress is shown in bytes.
  • Speaker identification, translators and language packs are offered where they are needed, with a Cancel that leaves your settings as they were.

Interface

  • A PRO or Free trial badge shows in the header, menu bar and About, and an ended trial is marked as ended.
  • Sound effects are off by default; the switch is in Simple mode too.
  • Improved Brazilian Portuguese wording in the app.

Fixed

  • A dictation take that ends early is completed automatically.
  • Short spoken terms such as 720p or 1080p are kept in the text.
  • A mistaken Clear or a change of file leaves a running file or subtitle job untouched.
  • The Manage License panel is back to its compact layout.
  • In the free version, choosing Subtitles shows the real Subtitles layout.

Improved

  • The memory read-out shows your Mac's real memory use, with DijiFlow's share marked.
  • When the system language pack would give a better translation, the app offers to download it before the job starts.
  • A finished subtitle or video job shows a completion card with the file details and a Show in Finder button.
  • Dictation is declined with a clear message when a file job leaves too little memory.
  • Buttons run the full width in every dialog.

v7.19

Major update2026

Subtitles, on-device translation and five language slots.

  • New Subtitles mode for files: a subtitle file, a video with a subtitle track, or the text in the picture.
  • Translate subtitles and dictation into 29 languages on your Mac, with nothing sent to a server.
  • Pro and the trial have five language slots, each with its own spoken and output languages.
  • Steadier subtitle timing and line layout. Chinese output in Simplified characters.
  • Long files use far less memory, even a feature length film.
  • A renewing subscription keeps Pro, days remaining count the paid term, and update notices say which way they point.
  • Stronger license and model protection, with no re-download and no change when you work without internet.

Subtitles for any audio or video file

  • Open the Transcribe a file tab, drop the file in, and set Output to Subtitles.
  • Pick the format: a subtitle file (.srt or .vtt), a video with a subtitle track, or the text burned into the picture.
  • Subtitles switch the model to Best, the one that keeps up with music and background noise. A smaller model prompts a warning.
  • Save, copy, or export the video. The export shows progress with a Stop button, and Show in Finder opens the result.
  • Save names carry the languages and the model, for example Meeting-en2tr-best-subtitles.srt.

Text in a language you did not speak

  • On the Output card, set Output language to something other than the spoken language.
  • The file is transcribed first, then translated, then the result card shows what you get.
  • With auto-detect, dictation recognises over 90 languages. Subtitles and translation cover 29, and each picker marks the weaker ones as beta or reduced accuracy.
  • Translation runs on your Mac. No audio and no text leaves the machine.

Five language slots, each with its own output language

  • Open the Slots tab. Pro and the trial have five slots, the free version has one.
  • Each slot carries a spoken language, a model, and an output language.
  • Give a slot an output language and dictation in that slot arrives translated, in whatever app you are typing in.
  • The overlay names the pair while you record, for example EN to TR.

Translation downloads only when you need it

  • The first time a language pair needs a download, the app fetches what it needs and shows a progress bar.
  • Nothing is downloaded for pairs your Mac can already translate.
  • Where a download is needed, the app picks the one your Mac can run comfortably.
  • You can remove a download later from the Models tab, in Advanced.

Subtitle timing that follows the speech

  • Lines are laid out over the speech, so a line does not open while nobody is speaking.
  • A long sentence splits at a clause instead of running past the phrase it belongs to.
  • Line width follows what a reader can take in at that speed, with a separate measure for Hindi.
  • Chinese comes out in Simplified characters throughout, whether it is the spoken language or the output language.

Long files, far less memory

  • A feature length film runs from start to finish without the memory climbing through the job.
  • There is nothing to switch on. It applies to every file you transcribe.

Shortcuts that recover after a permission change

  • If macOS switches Accessibility off for the app, typing everywhere on the Mac stays normal and the app tells you once.
  • Switch it back on in System Settings, Privacy and Security, Accessibility, and the shortcuts come back within a few seconds.

Update notices that say which way they point

  • Open the Settings tab and use Check for updates on the Engine card.
  • When the published release is newer than yours, the bar offers the update as before.
  • When it is older, or the two cannot be compared, the bar says the release is recommended and takes you to it.

Days remaining that count the paid term

  • The settings footer, the menu bar and the About window read from one figure.
  • The count runs to the end of the term you paid for.
  • If a renewal charge lands late, Pro is held and the app shows the hours of grace left.

License and trial state you can act on

  • A renewal is picked up before the term ends, so a renewing subscription keeps Pro.
  • Away from the network, Pro is held for up to 14 days instead of dropping to the free version.
  • A trial that has ended says so wherever it is mentioned, and each of those places offers Upgrade to Pro.
  • Status text and green and amber labels read clearly in light mode as well as dark.

Stronger license and model protection

  • License records and model files are checked more thoroughly before they are used.
  • Your models stay where they are through the upgrade, with no re-download.
  • Normal use is unchanged, and none of it needs the network.

v7.18

2026

Accuracy update, rebuilt vocabulary and a simpler Settings window.

  • Fewer phrases that were never spoken, fewer dropped words and fewer repeated words in English dictation, measured on more than 2,600 real recordings.
  • All 29 vocabulary packs rebuilt, with custom terms that match spoken phrases and spelled-out abbreviations.
  • Microphone switching is more reliable and your chosen microphone stays selected across restarts.
  • Redesigned Settings window: larger text, clear sections, choices as buttons. Simple view shows the essentials, Advanced shows everything.
  • Vocabulary: pick your languages from one list, each in a collapsible card.

Dictation accuracy, rebuilt and measured

  • How a recording is divided while you speak, and put back together afterwards, was rebuilt from the ground up.
  • Measured across more than 2,600 real English recordings: fewer phrases that were never spoken, fewer dropped words, fewer words repeated.
  • There is nothing to switch on. Every slot and every recording uses it.

Vocabulary rebuilt in all 29 languages

  • Open the Vocabulary tab and pick your languages from one list. Each one opens as its own card.
  • Up to three packs per language, shown on the card as the count you have selected.
  • Custom terms match a spoken phrase and its spelled out form, and fewer correct words are changed by mistake.

Simple and Advanced settings

  • The switch sits at the right end of the tab bar, and Simple is the starting point.
  • Simple carries the essentials: Slots, Settings, Vocabulary, Transcribe a file and Help.
  • Advanced adds the Models tab and the fine detail: recording sensitivity, the overlay, sounds, filters and storage.
  • The Engine card, Reset Settings and Uninstall are in both views. A control hidden in Simple keeps its value and keeps applying.

A settings window you can read

  • Text is larger throughout, and each card carries one category and one sentence of purpose.
  • Choices that exclude each other are buttons with a line of explanation under them, rather than small dots.
  • VoiceOver reads the title of each choice and says when one belongs to Pro.

Microphone reliability

  • The microphone you choose stays chosen across restarts, and switching devices is steadier.
  • The tip about USB microphones appears after your first finished dictation, rather than at launch.

Cleaner endings

  • A closing credit line that nobody spoke is removed from the end of a track before it reaches your text.
  • Turkish sign-off phrases and words repeated where two pieces of a recording meet are filtered out.
  • If saving a transcription or writing to history fails, the app tells you rather than passing over it.

v7.17

2026

Dictation shortcut reliability patch.

  • Your dictation shortcuts now keep working when macOS Secure Input gets stuck, with a clear notice while the condition lasts.
  • Performance and reliability fixes.

v7.16

2026

Reliability, clarity, and setup improvements.

  • Speech models now load without internet: everything they need ships inside the app.
  • Assigning a shortcut now explains itself: if a key is already in use by another slot, by macOS or by another utility, the app tells you where and what to do.
  • Setup picks a dictation shortcut that is actually free on your Mac.
  • Larger, higher-contrast text across the activation and setup screens.
  • Refined translations across the app's 29 languages.
  • Performance and reliability fixes.

v7.15

2026

Added a live dictation mode, plus vocabulary and reliability improvements.

  • New "Transcribe live" mode types text into the active app as you speak, character by character, instead of all at once when you release the key. Turn it on under "When to transcribe" in Settings — the default is still "Speak first."
  • Vocabulary correction is faster and recognizes specialized terms, brand names, and acronyms more reliably.
  • Fixed a rare case where silence or background noise could leave artifacts in the transcription.
  • Performance and reliability fixes.

v7.14

2026

Smarter vocabulary — multi-word terms and much stronger correction.

  • Custom and built-in vocabulary now support multi-word terms of up to eight words.
  • Split-up words ("digi flow"), close sound-alikes and — on Aggressive — acronym-style mis-hearings ("D.G. Flow") are corrected to your exact term.
  • Vocabulary terms always appear exactly as you typed them.
  • Redesigned Tips & Tricks in Help, with a ready-to-copy meeting-summary prompt.
  • Prices in the app now also show an estimate in your local currency.
  • Fairer plan changes, a drag-to-Applications installer and reliability fixes.

v7.13

2026

Our most polished file transcription yet.

  • Faster on Apple Silicon, accelerated by the Neural Engine.
  • A smoother progress bar with reliable time-remaining while a file transcribes.
  • A new completion popup with one-tap Copy and Save as file.
  • Choose exactly how many speakers to identify for cleaner speaker labels.
  • A clearer drag-and-drop overlay when you drop in a file.
  • More accurate timestamps and fewer repeated-line glitches on long recordings.
  • Transcripts now save as universal UTF-8 so they open cleanly in any editor.
  • General performance and reliability improvements.

v7.12

2026

Our fastest, clearest release yet.

  • Faster on-device transcription.
  • More accurate punctuation and capitalization in every language.
  • Fixed a rare case where very long recordings could finish without text.
  • General performance and reliability improvements.

v7.11

2026

Speed gains and stronger multilingual quality.

  • Quicker transcription, accelerated by Apple Silicon and the Neural Engine.
  • Noticeably better results in languages other than English.
  • More dependable transcription for large files.
  • Additional stability improvements.

v7.10

2026

Reliability and polish across the board.

  • Steadier results on very long recordings.
  • Cleaner, tidier output you can trust.
  • Many small interface refinements based on your feedback.
  • Faster, smoother day-to-day use on your Mac.

v7.0

2026

A major refresh — clearer plans and in-app account management.

  • A redesigned experience with Free, a 30-day Pro trial, and Pro.
  • View and manage your subscription right inside the app.
  • A friendlier first run that gets you set up fast.
  • Dozens of refinements throughout the app.

v6.5

2025

Higher-quality options for demanding work.

  • Choose higher-quality settings when you want the most accurate results.
  • Personalized model suggestions based on your Mac.
  • Quicker startup and a lighter footprint when idle.
  • More consistent results across languages.

v6.0

2025

Custom vocabulary that learns your world.

  • Teach the app the names, brands, and terms you use most.
  • Ready-made vocabulary sets for many languages.
  • Fewer corrections and more polished first drafts.
  • Reliable performance during long working sessions.

v5.5

2025

Speaker labels and timestamps for richer transcripts.

  • Automatically separate and label different speakers.
  • Add timestamps so you can jump straight to any moment.
  • More natural sentence breaks for easier reading.
  • Smoother handling of longer recordings.

v5.0

2025

Bring your existing recordings — file transcription arrives.

  • Drop in audio or video files and get clean, readable text.
  • Handy controls always within reach while you work.
  • Faster first-time setup so you can start in minutes.
  • Export your transcripts in a tidy, shareable format.

v4.5

2025

Better accuracy when the world is noisy.

  • Clearer results in busy rooms, cafés, and on calls.
  • Handles soft or fast speech more gracefully.
  • Smarter automatic capitalization and spacing.
  • A smaller download that installs in seconds.

v4.0

2025

A faster, more private foundation for everyday dictation.

  • Dictate into any app with a single shortcut — your words become text right on your Mac.
  • A cleaner, calmer interface that stays out of your way.
  • Noticeably quicker results, even on everyday Macs.
  • Wider language coverage with more natural punctuation.

Windows

Windows 10 / 11 (64-bit) · 8 GB memory

v7.19.3

Latest2026

More complete transcripts and subtitles, and cleaner subtitle lines.

  • Fixed rare cases where a passage of a file transcript could be dropped in quiet audio or around music.
  • Long dictations and long files are checked for stretches that came back too short, and completed.
  • Fixed a rare case where subtitles could drop words between two passages.
  • Improved subtitle line breaks, so a single word is not left on a line of its own.
  • Auto-detect identifies the spoken language of each recording.
  • New: transcribe or subtitle many files in one go. Pick several files and the app works through them one by one, showing which file it is on. Transcripts are joined into one text, subtitles are saved per file.

Transcripts and dictation

  • Fixed a rare case where a passage of a file transcript could be dropped when the audio was quiet.
  • Fixed a rare case where music, applause, or singing could remove the speech around it from a transcript.
  • In long dictations and long files, a stretch that came back too short is checked again and completed.
  • With automatic language detection, a completed passage is written in the language of the recording.
  • With Auto-detect as the spoken language, a recording is transcribed in the language that was spoken.
  • Files added together now run in name order, and you can sort them or drag them into the order you want.
  • Several files can be combined into one recording with one set of speakers, or saved as separate files to a folder you choose.

Subtitles

  • Fixed a rare case where subtitles could drop words between two passages.
  • Improved line breaks, so a single word is not left on a line of its own.
  • An opening word such as "Now," is kept with its sentence.
  • In English subtitles, word pairs such as "so that" and "and then" are kept together.
  • Subtitle translation into Japanese, Korean, Ukrainian, Finnish and Hungarian keeps each line with its own sentence.
  • Fixed rare cases where a translated subtitle line could stay untranslated or come back in the wrong language.
  • Translated subtitles no longer start with stray list numbers or letters, and a sentence that repeats in a file is translated the same way each time.

Fixed

  • Fixed the dictation overlay flickering when several takes overlapped.
  • Fixed a rare case where a few words could appear twice in a long recording.

Dictation

  • Dictated text is delivered reliably even when the key for the next slot is already held down.
  • When the output language cannot be changed, the reason is shown under the picker.

v7.19.2

2026

Reliable microphone selection, faster launch, and clearer memory handling.

  • Fixed a case where choosing a microphone while another one was connected could end dictation with a recording error.
  • The microphone list is limited to microphones you can dictate with.
  • Bluetooth microphones get the time they need to start.
  • The app is responsive right after launch, and the file window opens without delay.
  • When a model does not fit in memory, the app explains why and offers Dynamic Memory Optimization or a smaller model.
  • Fixed rare cases where the last sentence of a dictation, or a short dictation with digits, could be dropped.
  • Fixed a rare case where a sentence spoken after a pause could be left out.

v7.19.1

2026

Maintenance update for 7.19 on Windows: clearer progress, simpler names, and subtitles that are easier to read.

  • Burned-in subtitles are easier to read at any video size, including small clips.
  • File transcription shows a progress bar for each step, with a time estimate based on measured speed.
  • Choosing an output language checks that its translator or language pack is downloaded, and offers it before the run starts.
  • Speaker identification is available for subtitles, and the model can be downloaded from the file tab.
  • Models and translation engines are described by what they do.
  • Improved contrast in dark mode.

Fixed

  • A dictation take that ends early is completed automatically.
  • Short spoken terms such as 720p or 1080p are kept in the text.
  • A mistaken Clear or a change of file leaves a running file or subtitle job untouched.
  • The Manage License panel is back to its compact layout.
  • In the free version, choosing Subtitles shows the real Subtitles layout.

Improved

  • The memory read-out shows your PC's real memory use, with DijiFlow's share marked.
  • When the system language pack would give a better translation, the app offers to download it before the job starts.
  • A finished subtitle or video job shows a completion card with the file details and a Show in folder button.
  • Dictation is declined with a clear message when a file job leaves too little memory.
  • Buttons run the full width in every dialog.

v7.19

Major update2026

Subtitles, on-device translation and five language slots, now on Windows.

  • Subtitles mode for files: a subtitle file, a video with a subtitle track, or the text in the picture.
  • Translate subtitles and dictation into 29 languages on your PC, with nothing sent to a server.
  • Pro and the trial get five language slots, each with its own spoken and output languages.
  • Steadier subtitle timing and line layout. Chinese output in Simplified characters.
  • A renewing subscription keeps Pro, days remaining count the paid term, and update notices say which way they point.

Subtitles for any audio or video file

  • Transcribe a file and choose subtitles as the output.
  • Take them as a subtitle file (.srt or .vtt), as a video with a subtitle track, or burned into the picture.
  • Subtitles use the Best model, the one that keeps up with music and background noise.
  • The video export reports progress and can be stopped.

Text in a language you did not speak

  • Choose an output language other than the one that was spoken, and the file comes back translated.
  • With auto-detect, dictation recognises over 90 languages. Subtitles and translation cover 29, and the weaker ones are marked beta or reduced accuracy.
  • Translation runs on your PC. No audio and no text leaves the machine.

Five language slots, each with its own output language

  • Pro and the trial have five slots, the free version has one.
  • Each slot carries a spoken language, a model, and an output language.
  • Give a slot an output language and dictation in that slot arrives translated, in whatever app you are typing in.

Translation downloads only when you need it

  • The first time a language pair needs a download, the app fetches what it needs and shows a progress bar.
  • Nothing is downloaded for pairs your PC can already translate.
  • Where a download is needed, the app picks the one your PC can run comfortably.

Subtitle timing that follows the speech

  • Lines are laid out over the speech, so a line does not open while nobody is speaking.
  • A long sentence splits at a clause instead of running past the phrase it belongs to.
  • Line width follows what a reader can take in at that speed.
  • Chinese comes out in Simplified characters throughout, whether it is the spoken language or the output language.

Update notices that say which way they point

  • When the published release is newer than yours, the app offers the update as before.
  • When it is older, or the two cannot be compared, the app says the release is recommended and takes you to it.

Days remaining that count the paid term

  • The days you see run to the end of the term you paid for.
  • If a renewal charge lands late, Pro is held and the hours of grace left are shown.

License and trial state you can act on

  • A renewal is picked up before the term ends, so a renewing subscription keeps Pro.
  • Away from the network, Pro is held for up to 14 days instead of dropping to the free version.
  • A trial that has ended says so, and offers Upgrade to Pro.

Stronger license and model protection

  • License records and model files are checked more thoroughly before they are used.
  • Your models stay where they are through the upgrade, with no re-download.
  • Normal use is unchanged, and none of it needs the network.

v7.18

2026

Feature parity with the Mac release, plus Windows reliability fixes.

  • All 29 vocabulary packs rebuilt, with custom terms that match spoken phrases and spelled-out abbreviations.
  • Endings are cleaner: an unspoken phrase at the end of a recording is fixed, and long recordings split at natural pauses.
  • USB microphones and audio that could stop responding are fixed, and your chosen microphone stays selected.
  • Speed your PC gains after you install the app is picked up on its own, and the readouts are accurate.
  • Redesigned Settings window: larger text, clear sections, choices as buttons. Simple shows the essentials, Advanced shows everything.

Vocabulary rebuilt in all 29 languages

  • All 29 packs are rebuilt.
  • Custom terms match a spoken phrase and its spelled out form, and fewer correct words are changed by mistake.

Simple and Advanced settings

  • Simple carries the essentials, Advanced adds everything else.
  • A control hidden in Simple keeps its value and keeps applying.

A settings window you can read

  • Text is larger throughout, and each section carries one category and one sentence of purpose.
  • Choices that exclude each other are buttons with a line of explanation under them, rather than small dots.
  • Menus and layout mirror correctly for Arabic and Hebrew.

Cleaner endings and longer recordings

  • Words that were never spoken could appear at the end of a recording in rare cases. That is fixed.
  • Long recordings split at natural pauses more accurately.

Microphone and audio reliability

  • USB microphones and audio that could stop responding are fixed, and the microphone you choose stays chosen.

Faster transcription, picked up on its own

  • If your PC gains what it needs to run faster after you install the app, the app starts using it, with no reinstall.
  • What the app reports about speed and memory matches what is really happening.

v7.16

2026

Vocabulary, dictation reliability, and translation improvements.

  • Vocabulary correction now recognizes Thai and Hindi terms more reliably.
  • Fixed a rare case where a short phrase inside a long recording could be missed.
  • Fixed a rare case where transcribed text could be lost when the clipboard was briefly unavailable.
  • Long recordings now split at natural pauses more accurately.
  • More reliable graphics card detection on PCs with two GPUs.
  • Refined translations across the app's 29 languages.
  • Performance and reliability fixes.

v7.15

2026

Added a live dictation mode, plus vocabulary, speed, and reliability improvements.

  • New "Transcribe live" mode types text into the active app as you speak, character by character, instead of all at once when you release the key. Turn it on under "When to transcribe" in Settings — the default is still "Speak first."
  • Vocabulary correction is faster and recognizes specialized terms, brand names, and acronyms more reliably.
  • Major dictation speed improvement on CPU — the audio encoder now scales to the length of your speech.
  • Fixed a rare case where silence or background noise could leave artifacts in the transcription.
  • Performance and reliability fixes.

v7.14

2026

Smarter vocabulary — multi-word terms and much stronger correction.

  • Custom and built-in vocabulary now support multi-word terms of up to eight words.
  • Split-up words ("digi flow"), close sound-alikes and — on Aggressive — acronym-style mis-hearings ("D.G. Flow") are corrected to your exact term.
  • Vocabulary terms always appear exactly as you typed them.
  • Redesigned Tips & Tricks in Help, with a ready-to-copy meeting-summary prompt.
  • Prices in the app now also show an estimate in your local currency.
  • Fairer plan changes and reliability fixes.

v7.13

2026

Our most polished file transcription yet — now with broader GPU support.

  • Faster transcription across NVIDIA (CUDA), AMD, and Intel GPUs, with smarter automatic fallback.
  • Improved speaker identification for cleaner, more reliable speaker labels.
  • A smoother progress bar with reliable time-remaining while a file transcribes.
  • A new completion popup with one-tap Copy and Save as file.
  • Choose exactly how many speakers to identify for cleaner speaker labels.
  • A clearer drag-and-drop overlay when you drop in a file.
  • More accurate timestamps and fewer repeated-line glitches on long recordings.
  • General performance and reliability improvements.

v7.12

2026

Our fastest, clearest release yet.

  • A simpler Windows setup that helps you pick the right download for your graphics card.
  • Faster on-device transcription.
  • More accurate punctuation and capitalization in every language.
  • Fixed a rare case where very long recordings could finish without text.
  • General performance and reliability improvements.

v7.11

2026

Speed gains and stronger multilingual quality.

  • Quicker transcription, with GPU acceleration across NVIDIA, AMD, and Intel.
  • Noticeably better results in languages other than English.
  • More dependable transcription for large files.
  • Additional stability improvements.

v7.10

2026

Reliability and polish across the board.

  • Steadier results on very long recordings.
  • Cleaner, tidier output you can trust.
  • Many small interface refinements based on your feedback.
  • Faster, smoother day-to-day use on your PC.

v7.0

2026

A major refresh — clearer plans and in-app account management.

  • A redesigned experience with Free, a 30-day Pro trial, and Pro.
  • View and manage your subscription right inside the app.
  • A friendlier first run that gets you set up fast.
  • Dozens of refinements throughout the app.

v6.5

2025

Higher-quality options for demanding work.

  • Choose higher-quality settings when you want the most accurate results.
  • Personalized model suggestions based on your PC.
  • Quicker startup and a lighter footprint when idle.
  • More consistent results across languages.

v6.0

2025

Custom vocabulary that learns your world.

  • Teach the app the names, brands, and terms you use most.
  • Ready-made vocabulary sets for many languages.
  • Fewer corrections and more polished first drafts.
  • Reliable performance during long working sessions.

v5.5

2025

Speaker labels and timestamps for richer transcripts.

  • Automatically separate and label different speakers.
  • Add timestamps so you can jump straight to any moment.
  • More natural sentence breaks for easier reading.
  • Smoother handling of longer recordings.

v5.0

2025

Bring your existing recordings — file transcription arrives.

  • Drop in audio or video files and get clean, readable text.
  • Handy controls always within reach while you work.
  • Faster first-time setup so you can start in minutes.
  • Export your transcripts in a tidy, shareable format.

v4.5

2025

Better accuracy when the world is noisy.

  • Clearer results in busy rooms, cafés, and on calls.
  • Handles soft or fast speech more gracefully.
  • Smarter automatic capitalization and spacing.
  • A smaller download that installs in seconds.

v4.0

2025

A faster, more private foundation for everyday dictation.

  • Dictate into any app with a single shortcut — your words become text right on your PC.
  • A cleaner, calmer interface that stays out of your way.
  • Noticeably quicker results, even on everyday PCs.
  • Wider language coverage with more natural punctuation.

Download the latest version