12.13

Improvements:

• The app gives a clearer message about needing to update to macOS 14.4 or newer to use the Parakeet models
• Improved stability of the dictation feature
• Added an extra warning to prevent data loss by not transcribing app audio recordings
• The dictation overlay is presented above third party spotlight-like overlays
• Dictation works again on www.chatgpt.com
• Design tweaks
• Font size can be changed with Cmd and the -/+ buttons instead of just the -/+ buttons
• WhisperKit models are presented for m1 users again

Bugfixes:

• Fixed an issue where the app could say "Whisper is busy"
• Fixed an issue where batch transcriptions for short audio files with Parakeet models could fail
• When modifying a custom dictation AI prompt it's enabled without having to toggle it first

12.12.2

New:

• Added support for the new Parakeet model, which is extremely fast and accurate! It can transcribe at up to 300 times real speed on the latest Macs. This current release only supports English, please try it out and let us know what you think! (Pro)

Improvements:

• The progress indicator now shows progress more accurately and calculates the time remaining more often
• Design cleanup around the app
• More localization of error states
• Added support to meeting detection for Zen and Dia Browser

Bugfixes:

• Fixed an issue where recordings could fail on Macs that have their language set to Portuguese due to a file name issue
• Fixes for using the Claude 4.0 models

12.11

New:

• Batch transcriptions can now be paused and resumed (transcriptions will pause after the active item is finished) (Pro)
• Added support for the new Claude 4.0 models (Pro)
• You can now drag out .whisper files from the sidebar to other apps on your Mac

Improvements:

• Meeting notifications don't require two clicks anymore (one to activate the window, one to click the button)
• Pro users no longer briefly see pro badges on launch
• Dictation can now also be muted when playing audio through a Studio Display
• Improved the new floating video player
• Clearer error messages when LM Studio models return errors

Bugfixes:

• No longer showing an empty window in Mission Control if you use the menubar only mode
• Fixed an issue where the app was not immediately responsive on launch
• When pausing with spacebar, the videoplayer now also pauses correctly

12.10

New:

• Added a new floating video player mode

12.9.1

New:

• You can now select how many speakers are in the transcript and run speaker recognition again for improved results (only when using local WhisperKit models)
• Pro models can now be used for dictation for free users as well

Improvements:

• Improved the remove dashes enhancement feature to include more occurences of dashes at the end of segments as well as only segments with a dash
• Added the option to only export favorited segments to segment or subtitle export
• Added more options for MDM deployments

Bugfixes:

• Fixed a crash when deleting the last segment on the speaker view
• Fixed an issue where meeting recordings could have sped up audio

12.8

New:

• Automatically transcribe files in watch folders! Files that are added to your watch folders can now automatically be transcribed into multiple formats. Thanks for the feedback and testing on this!

Improvements:

• YouTube downloads are working again!
• Added more options for MDM deployments to set or limit certain features
• Added back the ability to mute audio while dictating. MacOS 15.4 removed the option to pause audio, so this brings back that functionality in the form of muting audio instead.

Bugfixes:

• Fixed an issue where the app window would sometimes not open when clicking the Dock icon
• Fixed an issue where dictation would not start unless the main window was opened when using menu bar only mode
• Fixed an issue where dictation errors could not be dismissed

12.7

Improvements:

• You can now rename files from the sidebar
• Fixed an issue in the dictation onboarding
• Removed timeouts for ElevenLabs and Deepgram for when you're transcribing long files
• Improved support for more .ogg files such as WhatsApp audio messages
• Updated the design of the task overlays in the bottom right corner of the main window

Bugfixes:

• Fixes for recording meetings in a Chrome PWA
• Fixed a visual bug that could appear at 100% when using cloud transcription

12.6

Improvements:

• Improved the first run experience for new users
• Whitespaces are automatically stripped when adding your own API keys
• The history sidebar looks nicer

Bugfixes:

• Removed "Pause Audio during Dictation" since macOS 15.4 broke it. We're working on bringing it back.

12.5

New:

• Punctuation in dictation. You can now toggle punctuation mode which will automatically add punctuation such as "new line", "question mark" and others.

Improvements:

• Improvements to meeting recording
• Fixed an issue where YouTube downloads were using the wrong audio track
• Speaker recognition now also works in System App Audio recordings and when downloading URLs
• When using the ElevenLabs API the file size limit is now set correctly
• Improvements to speaker recognition
• New lines are stripped out of dictations that use Gemini
• Meeting recordings can now be transcribed with cloud providers as well as long as the file is under the file limit for the provider
• Added an export option to only export favorited segments
• Added support for the new GPT 4.1 models

12.4

Improvements:

• You can now rename meetings from the sidebar (right click) or the toolbar
• Clearer error codes when a meeting recording can not be started
• Added a space after each dictated piece of text
• Text is now aligned to the leading side of the view instead of justified
• Added an icon for the history bar for system audio recordings
• Added the option to show/hide the speaker names in the segment view
• You can now export the combined audio from a meeting
• Added support for recording meetings in Orion browser
• Stability improvements

Bugfixes:

• Fixed a potential crash when recording a meeting
• Fixed an issue where duplicate files would get saved which would take up a lot of storage when editing a whisper file
• Fixes for Deepgram integration
• Fixed an issue where error alerts in dictation mode could not be dismissed

12.3

Improvements:

• Improvements to the record meeting feature to make it more stable.
• App Audio recordings now show the app icon of the app that was recorded
• When you change the AI prompt in the dictation popup, that will be the default from that point on
• Added a space after a dictated sentence
• Improvements to speaker recognition
• The onboarding is improved for MDM deployed versions of the app

Bugfixes:

• Fixed an issue where meeting recording could crash
• Fixed an issue where a microphone would not be recorded in meetings in specific setups

12.2

New:

• You can now filter your transcript by unknown speakers, and quickly jump to the next occurence of an unknown speaker
• Added the option to use custom cloud transcription providers based on the OpenAI whisper spec. This can be used for running transcription on your own private server endpoints.
• Added support for language specific models. Currently Swedish and Japanese, more are coming.

Improvements:

• Speaker recognition now also works for meetings and batch transcriptions
• Added support for o1 and o3 models for the OpenAI AI service
• Added a toggle for microphone recordings to enable speaker recognition in case you're recording multiple people at the same time
• Added a retry button for when dictation fails with a cloud provider

12.1.1

New:

• Support for more models for Deepgram, including their model specifically trained for the medical field
• You can now choose to automatically save each transcript as a .whisper file. This will be the default behaviour soon, but you can already enable it from settings.

Improvements:

• You can now undo accidentally deleting a speaker from the sidebar
• Improved dictation with AI prompts so that the AI service does not reply to your dictation
• Highlighted search words are improved in the transcript view
• Added the option to also delete a file when you want to remove it from your history list

Bugfixes:

• Fixed an issue where dictation could ask for an AI service even if it was not set up

12.0.1

New:

• Automatic Speaker Recognition! Finally! Automatically recognise speakers in your recordings using local models. To use it, make sure you select a model that supports speaker recognition (WhisperKit). After your transcription is complete it will automatically be grouped by speaker. We're still working on improvements so let us know what you think! (Pro)
• Click on segments in the transcript view to start playback from there
• Play the first segment for an identified speaker from the sidebar to make it easy to identify which speaker is who
• Added the option to automatically improve spelling and grammar for dictations without having to use a prompt (Pro)

Improvements:

• You can now adjust the speaker for a paragraph from the transcript view
• Assign segments to a different speaker using the keyboard shortcuts (1,2,3 etc)
• You can now use cloud transcription models in the app without having to first download a local model
• You can now reassign all segments from one speaker to another one
• Speaker recognition now also works for M1 users
• Added a badge to identify which models support speaker recognition
• Made it clearer when the app is identifying speakers instead of it appearing like progress is stuck at 100%
• Small design tweaks and bugfixes
• Improved the design for prompts in settings
• Speaker recognition is now also enabled for microphone recordings

Bugfixes:

• Fixed an issue where extra spaces were added for some languages such as Thai and Chinese

11.13

New:

• Added the option to show timestamps for grouped speaker paragraphs in the transcript view

Improvements:

• New design for the speakers in the sidebar
• Improved grouping of speakers by removing the 'Speaker 0' issue. More improvements are on the way.
• Added some more empty state screens

Bugfixes:

• Fixed an issue where the font size of the transcript view could not be adjusted

11.12

New:

• First release with speaker recognition for local models! Try it out with a WhisperKit models (Pro only and M2 or newer only for now). Please send us your feedback on what we should improve.
• New display mode where the transcript is grouped by speaker paragraphs. Only visible if multiple speakers are available.

Improved:

• Improved performance while fast loading models

Bugfixes:

• We're working on solving an issue on the latest macOS 15.3 in combination with M1 series Macs. WhisperKit models are temporarily not recommended for M1 series users while we figure out a good path forward.

11.11

Improvements:

• Transcription complete notification are no longer shown when the app is in focus
• Improved the performance of the markdown scroll view on the export page
• Fixes a bunch of export preview edge cases
• Improved the PDF export design
• You can now hit return when searching to move to the next instance

Bugfixes:

• If translation to a language fails for whatever reason it's no longer added to the recent list
• If a WhisperKit model crashes the app will now turn off fast loading to prevent it from happening again while we wait for Apple to fix this bug in a new version of macOS

11.10.2

Bugfixes:

• Deepgram can now be used for dictation mode again
• Users who had their WhisperKit settings set to use the GPU are now properly reset to Neural Engine if they run into a macOS related crash
• Small bugfix that nobody will notice

11.10

New:

• Automatic Speaker Recognition can now be used with Deepgram and ElevenLabs cloud transcription. Local support coming very soon!
• Added support for OpenRouter AI services
• Added support for Deepgram Nova for cloud transcription
• Added support for ElevenLabs Scribe for cloud transcription
• Add support for ChatGPT 4.5

Improvements:

• Improved the design and spacing in the pdf export
• Automatically select the newly created Find and Replace item in settings
• Settings loads faster the first time you open it
• Adding a new prompt in settings is now faster since the sheet will appear immediately
• Added a toggle to disable sending of anonymous telemetry data
• New notification when a transcription made with a Whisper C++ causes a lot of repetitions
• Fixed an issue where loading a WhisperKit model would crash with Fast loading enabled. If you get the message again, it should not happen afterwards.
• Added a button to retry sending your dictation to a cloud provider or AI service if something went wrong
• You can now move to the next found search result by hitting return
• Improved the design of search highlighting

11.9

New:

• Added initial support for Dvorak (and other) keyboards for dictation. Enable it from Settings > Advanced and let us know if you run into anything if you have an 'exotic' keyboard setup

Improvements:

• The meeting detected notifications now automatically dismiss after ten seconds
• Improved model selection flows in edge cases
• Fixed an issue where dictation could crash with specific hardware
• Clearer errors when a file can not be transcribed in batch mode
• You can now switch between found search text results with ⌘G and ⌘⇧G
• Improvements to translating with Apple Translation
• Added support for Claude 3.7 Sonnet

Bugfixes:

• Fixed an issue where you could not remove items from the batch list

11.8

New:

• You can now choose when transcription finished notifications should be triggered based on the duration time to finish the transcript
• Added the option to export as JSON (Pro)

Improvements:

• Improved textfield detection in dictation (better support for Sublime and other text editors)
• Improved the contrast of the start recording button on global in dark mode
• Added a correct minimum width for the sidebar in settings
• Hitting return during an active dictation no longer finishes the dictation

Bugfixes:

• Fixed an issue where when you end a meeting the notification in the app could show Zoom instead of the app you were recording

11.7

New:

• Huge speed and stability improvements for the dictation feature. Long dictations now appear almost instantly because your words are transcribed in the background as you're talking. Try it out and let us know what you think!
• Added meeting detection for Edge
• When scrolling through the playback bar, the transcript segment is highlighted
• You can now name your microphone recordings

Improvements:

• Improved loading of recorded Teams files when using WhisperKit models
• Added support for the latest Gemini models
• Improved meeting detection for Skype
• Added a clearer alert when trying to use a cloud provider to transcribe a meeting since the file size limits don't allow it currently
• You can enable the dictation feature to work everywhere, not just when a text field is detected. To turn it on, go to Settings > Advanced.
• Clearer errors when LM Studio returns an error response
• When during dictation no textfield is focused, your dictation will be added to your clipboard
• Support diacritics in HTML export
• You can now remove previously detected, but unused, microphones from the microphone priority list
• Fixed an issue where the sidebar would jump when opening settings for the first time (finally!). If you still see it, let us know.

Bugfixes:

• Fixed an issue where you could not switch AI prompts easily from the dictation bubble

11.6

New:

• Free users can now use all models! See how much the quality improves when using the largest models. You won't be able to copy, export or otherwise use the transcript unless you upgrade to Pro.
• Added the option to unload models after a fixed number of minutes. This can be useful for users with lower amounts of RAM. Models are loaded back into memory when starting a new transcription.
• You can now choose to pause and automatically resume playing media when using the dictation feature
• Active downloads now show in the overlay on the homescreen
• Dictation Word Dictionary: Add words and terms that the dictation feature misinterprets. When an AI prompt is active during dictation, these words will be corrected automatically.

Improvements:

• When transcribing meetings and other recordings with multiple recordings, the progress bar now shows more clearly how many files are remaining
• Fixed the model and language picker designs in the sidebar and other places
• Models are now loaded on first transcription instead of on the homescreen
• The Assistant sidebar is no longer cropped when the window is very small
• Speed improvements all across the app, it should feel a lot snappier!
• Improvements to active meeting detection
• Speaker colors that get assigned in meeting recordings are more consistent

Bugfixes:

• Fixed an issue where the "Manage Models" button could flash on launch
• Fixed an issue where the main app could not be opened from the global overlay if the main window was closed before

11.5

New:

• Improved support for .oga files
• Added support for more automatic meeting recording applications such as Amazon Chime, Skype, Discord, WhatsApp and web browsers
• Automatic Meeting Detection now supports a lot more apps. Please let us know if you run into any issues

Improvements:

• The summary feature now takes into account the original language of the transcript and will output in that same language by default
• You can now choose the default name for you and other attendees in meetings
• Fixed an issue where YouTube links could not be transcribed
• Improved dictation performance to where it should not answer your dictations if you don't specify it as a custom prompt
• Added a toggle to enable or disable automatic summarisation when you open the summary view
• You can no dismiss detected meeting overlays
• Better error handling in certain scenarios
• Improved meeting detection in Zoom and Webex
• Remove mentions of BLANK_AUDIO from empty dictations

Bugfixes:

• Fixed an issue where you could not cancel a YouTube download

11.4.3

Improvements:

• Clearer error messages when rate limited

Bugfixes:

• Fixed an issue that could cause the app to hang.
• Fixed a crash that could happen when using automatic meeting notifications

11.4

New:

• You can now choose to save Global recordings in your history for easy access later (thanks Martin)
• You can now easily remove segments that only include words with asterisks surrounding it
• Added support for Deepseek as an AI service

Improvements:

• Improved the design of the Record System Audio screen. It's now simpler to select which app you want to record and it's clearer when the microphone is recorded as well.
• Added a recording time indicator for the Record System Audio feature
• The app now remembers if you wanted to record your microphone during app audio recordings
• When adding new speakers, their default color should no longer clash with existing speakers if possible
• If no model name is selected for a Ollama or LM Studio service the title will show correctly
• You can more easily add a new speaker by right clicking a segment, even if there already are speakers available
• The speaker percentage view now properly adds up to 100%

11.3.2

Bugfixes:

• Fixed an issue where a meeting notification could show the wrong meeting app
• Improved detection of Teams meetings
• Meeting app notifications are now displayed in middle top of the screen
• When adding new speakers the color should be unique
• Fixed an issue where the main app could not be opened from the menubar or the menubar would not show up.

11.3

New:

• Automatic Meeting Detection! The app will now detect if you are in an active Zoom, Teams or Webex meeting and will notify you to automatically record the meeting. Free while in beta. (Pro)
• Start recording a meeting straight from the menubar
• The history sidebar is now grouped by days and weeks
• You can now use app specific AI prompts with the dictation feature. The app can automatically switch them for you so you can use a different prompt while answering emails or while coding for example. (Pro)

Improvements:

• Improved the design of the AI services screen in settings
• Improved the design for the speakers section in the transcription sidebar
• Speakers are added to the sidebar when recording meetings and podcasts again
• The audio files from your microphone and the rest of the meeting participants are now saved together and are accessible from the history sidebar
• Active meeting recordings are shown in the active task area in the bottom right of the main app
• You can now record multiple meetings and then transcribe them afterwards
• Fixed a flicker on the dictation settings screen
• Left clicking the menubar now shows the menu, while right clicking opens Global
• Added a button to check for updates and to open settings from the menubar
• You can add new speakers from a right click on a segment again
• Added info per speaker to the information sidebar. You can see words, characters and percentage of words spoken per speaker.
• Improved how the app handles when it's only active in the menubar
• You can now favorite, delete or assign speakers to a segment that you're hovering over, without having to select it first

Bugfixes:

• Speakers are added in the sidebar when transcribing podcasts and system app audio again
• Fixed a crash when using VAD with WhisperKit
• Fixed an issue where you could not use AI features after opening a Global transcription in the main app
• Fixed an issue where audio would not playback for podcasts or meetings until you saved
• Fixed a potential crash related to a corrupt wav file

11.2

New:

• Added a search bar in settings to more easily find all available options
• New design for settings, please send us feedback on what you think!

Improvements:

• Added support for the new Gemini 2.0 flash experimental model
• The name for downloaded files is prettier
• Added the option to show the full transcript on the Assistant tab
• When selecting a prompt from the sidebar in Assistant Chat it is no longer automatically sent so you can adjust it
• Cleaned up the homescreen a little bit
• Added icons in the model picker on the homescreen

Bugfixes:

• The Assistant tab now remembers if you last used the Chat or Summarize feature
• The release notes message in the bottom left won't show on first launch
• HTML previews now load correctly
• Fixed an issue where text would blink while transcribing when using WhisperKit with VAD
• Various small bug fixes around the app

11.1

New:

• The search bar now shows how many matching words were found in your transcript
• Added the option to use Voice Activity Detection for WhisperKit models. This will increase your transcription speed and will remove issues related to empty chunks of audio. Try it from Settings > Advanced (Pro)
• You can now choose to only show the app in the Dock, the menubar or both
• Transcripts created with WhisperKit models will now highlight individual words during playback on the transcripts view (Pro)

Improvements:

• The html export now changes background and text colors based on light and dark mode
• Dictations will no longer show up in clipboard history managers such as Alfred
• Added a button to create a new folder when choosing your save location
• The videoplayer playback speed matches the audio playback if you increase or decrease the speed
• If a Whisper model can not be loaded on launch, the next best model will be loaded
• You can now increase and decrease the font size with ⌘- and ⌘+
• You can increase or decrease the playback speed of the player with the < and > buttons on your keyboard
• The copy button now takes into account the display mode you have selected
• Whisper files show a nicer filename
• The export preview text now shows your entire transcript
• When translating you will see a "Translating..." indicator
• The sidebar now animates nicely when switching between full and compact mode
• You can now add timestamps and speaker names when using AI features.
• You can now remove translations by right clicking them in the sidebar
• Added quick options for combining segments to sentences and for removing "- " at the start of segments

Bugfixes:

• You can't click the "Start" button in global mode anymore when no models are loaded
• When starting playback by clicking on a segment, the videoplayer will now be sync

11.0.1

Bugfixes:

• After exporting, the MacWhisper window doesnt disappear any longer.
• Fixed an issue where it might not be possible to setup Dictation using a shortcut that was already configured previously.

11.0

New:

• Completely new design for the transcript view, with a convenient sidebar for easy access to the most used features
• Adjust the font size on the transcripts and segments views from tiny to very large
• Collapse the sidebar for a focused view of your transcript
• You can now choose to show padding around your transcript for a cleaner view
• Speakers are now added on a per transcript basis and can be added more easily from the sidebar
• Added a clearer Pro overview screen in settings
• Improved transcript view design with flexible sidebar
• Add speakers directly in the transcript view
• View information about the current transcription on the new Info tab
• You can now choose to use the right option key for dictation

Improvements:

• When retranscribing with a different model the homescreen won't flash anymore
• All sound effects are now at the appropriate volume (thanks Konstantin)
• You can now adjust the colors per speaker from the speaker sidebar section
• You can now assign speakers to a segment by using the 1,2,3... keys on your keyboard
• Segments now have a background color that matches the speaker associated with that segment
• Responses on the AI screen will stay visible when switching display modes
• Improved the design of the AI Services view in settings
• Faster performance of the preview on the export view
• Added a copy button to the toolbar for easier access

10.9.2

Improvements:

• Added a "Don't show this again" option to more dialogs. These can be managed via the "Show Save Confirmation" preference in Settings.

Bugfixes:

• Fixed an issue where the "Don't show this again" dialog preference might not be saved correctly.

⚠️ Last Ventura Update

• This is the last update that supports macOS 13.0 (Ventura). Please update your Mac to 14.0 or higher to use new features we are adding to MacWhisper. If you run into issues on this version on Ventura please let us know and we'll try our best to fix them so that the app is as stable as can be for users on 13.0.

10.9.1

Improvements:

• Fixed an issue where Microsoft Teams recordings would sometimes fail to load
• Added a clearer "Do not ask again" button to the save transcription overlay
• Fixed an issue where dictation would paste clipboard content
• Improved support for detecting textfields in apps
• Dictation now works with ChatGPT, Anthropic and other overlays
• Video player content is now in sync with audio

Bugfixes:

• Fixed an issue where release notes would be shown in the wrong scenario
• No longer adds a file in the user's documents folder. It can be safely deleted after up

10.8.1

Bugfixes:

• Dictation: Fixed missing spaces bug when dictating into some browser fields

10.8

New:

• Dictation is up to 10x faster for longer chunks of text
• Added a setting to launch MacWhisper at login

Improvements:

• Improved the position of the dictation overlay on secondary displays
• Improved compatibility of the dictation shortcut keys
• The changelog is not shown to users who have not seen the onboarding yet
• Defaulted the WhisperKit settings to use the Apple Neural Engine
• Fixed an issue where the Global view keyboard shortcuts would stop working after opening it multiple times

Bugfixes:

• Fixed a crash when editing the last segment in a file

10.7

New:

• Custom keyboard shortcuts for dictation are back! Besides using the Fn or right Cmd key, you can again choose your own keyboard shortcuts to start and stop the dictation feature. Thanks for the feedback!

Improvements:

• Fixed an issue where some files could not be loaded with a WhisperKit model.

10.6

New:

• Watch Folders: Add folders that you want the app to observe, and whenever a new compatible file is added you can quickly transcribe it. Send us your feedback on how we can make it better for you! (Pro)

Improvements:

• You can choose to toggle Dictation instead of having to press and hold the dictation key. You can adjust this from Settings > Dictation.
• Updated to use the latest Claude 3.5 Sonnet model
• Your old dictation keyboard shortcut gets disabled after enabling the new dictation features
• Use the correct keyboard glyps in dictation settings
• Added extra options for MDM deployment to disable AI services and Cloud Transcription
• Faster performance when using WhisperKit models

10.5

New:

• Push to Talk Dictation! We've reworked the dictation experience to be faster and more convenient. Just press and hold one of the dictation buttons you choose, talk, and release to type in any textfield on your Mac. Enable it by clicking the Dictation button on the homescreen.
• Full Support for Writing Tools on macOS 15.1. Locally summarize, rewrite and improve your transcripts with Apple Intelligence
• Dictation and Global history, view your past 50 dictations and copy them to reuse
• Added support for Google Gemini AI models (Pro)

Improvements:

• Improved the textfields when adding your own AI prompts

Bugfixes:

• Fixed an issue where dictations could take a long time to complete
• Fixed an issue where transcriptions couldn't be saved when editing a segment
• Fixed an issue where the settings window would disappear when opening the app with it open

10.4

New:

• Added support for LM Studio for very fast local AI models
• Added support for the xAI API

Improvements:

• Removed the num_ctx parameter for Ollama, which should improve performance, let us know if you run into anything.
• Tweaked the design of the select model button on the AI page.
• The inline video player no longer shows up on the AI screen where it overlaps with the prompt view.
• Hide the "Recently Used Languages" section in translation settings if there are no languages yet
• Fixed a strange animation in the onboarding
• Improved the design of the AI services screen
• Show clearer error when you're trying to use the MacBook microphone with the lid closed

Bugfixes:

• Fixed "No context length could be determined" error for some Ollama models
• Fixed an issue where two of the same microphones could appear in the microphone priority list
• Fixed "Finished without any text" glitch when using Cloud transcription

10.3.1

New:

• Added support for using Groq as a Cloud transcription provider (Pro)
• You can now configure which microphone should be used for recordings. From Settings > Microphone you can choose 'System Default', 'Fixed Microphone' or 'Priority List'.
• Added a changelog screen for larger updates to highlight new features

Improvements:

• Find & Replace: If case sensitive is turned on, then the replacement should also be case sensitive.
• Find & Replace: make regex search pattern safe for special characters such as !, ? etc.
• Textfields look nicer in settings
• Improved button placement and design in settings
• Ollama models now have a higher token limit

Bugfixes:

• Fixed minor memory leak when undoing changes

10.2

New:

• Default Batch Export settings. You can now setup which formats should be used for batch transcriptions from Settings.

Improvements:

• Improved performance of batch transcription screen when transcribing more than 20 files in one go
• Design tweaks for the batch transcription view
• You can now use a custom OpenAI model as well (for anyone with access to gpt-5)

Bugfixes:

• Fixed an issue where the internet connection was checked too often for people whose internet connection dropped out

10.1

New:

• Add support for OpenAI hosteda on Azure
• The Global feature is now also available to free users

Improvements:

• Added a delete button to ai services
• Added clearer errors per AI service provider
• Clarified what urls are valid for custom endpoints
• New icon for custom AI services

Bugfixes:

• You can now add multiple Groq services and use different models for each
• Fixed an issue where full transcripts where exporting with timestamps
• Timestamps in segments exports are now displayed correctly

10.0.1

Improved:

• Small fixes and improvements

10.0

New:

• Added the new Whisper Turbo model which has the same accuracy as Large, but can transcribe at 20x realtime. Try it out!
• Local AI Models with Ollama support. You can now use any AI model that you run through Ollama on your Mac.
• Custom AI providers. You can now add your own custom AI providers which use the OpenAI API spec. Add them from the AI Services tab in settings and then use it across the app.
• Grog AI support. Use the Grog service to run AI prompts on your transcripts with your own API key.

Improved:

• Global and Dictation now also supports removing duplicates as well as your ignore/replace list.
• The Global window can't be dismissed while it's transcribing to make sure your recordings don't disappear accidentally.
• Cleaned up the model picker dropdown to more easily differentiate model capabilities and engines.
• Timestamps can be enabled again in segment exports.
• Global timestamp changes now show up correctly in exports.
• You can adjust timestamps correctly now on segments, even if they go to the next minute.
• The Manage Models screen is more resilient to a lost network connection.
• Extra checks to prevent WhisperKit models from showing up on Intel devices (where they are not supported).
• Added extra "Copy logs to clipboard" option to help debug issues.
• Improved the design for the AI Prompts page.
• Prompt titles and prompts now have a nicer sheet with more space for editing.
• Added descriptions for the different export options.
• Dictation now works correctly in Arc Browser

9.15

Improved:

• Fixed an issue where the app could crash on macOS Sequoia
• Fixed an issue where quick exports to full transcripts would include timestamps
• WhisperKit models are now shown in the manage models list by default as well. Try one of the out from the dropdown in the top right for even more accurate transcripts.
• Improved the UX around the Global mode. If you have stay on top enabled but have not started a recording yet the window will still dismiss.
• When you dismiss the Global mode while a recording is active, it will still be there when you open Global again
• Improved compatibility for more mp4 files

9.13

New:

• You can now easily re-transcribe using a different model or input language, directly from the Transcripts screen.
• The Segments screen now supports scrolling up and down using the keyboard while the text fields are focused.

Improved:

• Added a "You must first fund your OpenAI account to use this API key" message to the OpenAI Settings screen, if we detect the account has run out of credit.
• Migrated WhisperKit (beta) models out of the Documents folder. If model files are offloaded to iCloud and not available locally, the WhisperKit models will need to be re-downloaded in the Manage Models screen. Please note that the first load of these migrated models may take some time as they are optimized for your specific system configuration.
• Resolved an issue where OpenAI Cloud Transcriptions would not stop correctly if cancelled midway through the process.
• Fixed an issue where YouTube downloads would continue running after the download was cancelled.

9.12

Improved:

• More reliable "typing" output support for the Dictation feature, with fixes for specific apps such as rich text fields in Chrome and Firefox. Please email us at support@macwhisper.com if you still run into issues!
• File History list is now more resilient at opening files that have been moved.

9.11

New:

• Support for dictation recordings under 1 second has been added.

Improved:

• Typing dictation output is now delayed until all keyboard shortcut modifier keys are released.
• Enhanced error handling when dictation does not have microphone permissions.

9.10

New:

• You can now also use Anthropic as the dictation AI service.
• You can choose which model to use for dictation (in Settings), separately to the model you use on the AI Transcription screen.

Improved:

• When modifying a dictation prompt, the matching active prompt will now also pick up the change.
• Improved accuracy of dictation typing, and squashed a few bugs (for example: typing times into Apple Mail now works correctly).
• The main MacWhisper window no longer appears when dictation is activated.
• The default Dictation AI Prompts have been improved to deliver better responses from the LLM.
• WhisperKit models stream responses into the Transcript view. This release fixes a UI glitch.

9.9

Improved:

• Dictation: Fixed an issue where dictations could have spaces in wrong places
• Dictation: Fixed an issue where two spaces could turn into a period
• Global: Fixed an issue where the audio could not be saved when opened in the main app
• ChatGPT: Added support for "GPT-4o Latest" and "GPT-4o 64k Output Alpha" (this last one requires you have access to it before you can use it)

9.8

Improved:

• Fixed an issue with the YouTube downloader not working correctly
• When you have the 'Translate to English' feature enabled it will now be shown on the home screen

9.7.1

Improved:

• Dictation now also works in Mimestream and Microsoft Office apps. If you notice it does not work in a specific app let us know at support@macwhisper.com
• The default OpenAI model is now GPT4-o mini
• The dictation feature will now use the OpenAI model you have enabled, instead of always using GPT4o
• Fixed an issue where YouTube videos were not downloading because the 'High Quality' setting was turned on. We disabled the setting for now.

9.7

New:

• Auto Translations. You can now automatically translate transcripts in specific languages. Add the languages you want to translate, and what languages to translate them into from settings.

Improved:

• Added a toggle to Global Find and Replace to only match on complete words or also on parts of words. Disable this for languages that don't use spaces to separate words such as Chinese.
• The Global window now shows above all other apps even if you disable float on top
• When performing batch transcriptions, you will no longer see the transcripts happening in the background.

9.6

Improved:

• Added support for the new GPT4o 2024-08-06 model which is cheaper and has 16000 output tokens so it can be used for large summaries
• The AI response view now shows markdown text properly
• Small fixes around dictation and prompts

9.5.1

Improved:

• Added a toggle to show Global on top of all other apps and keep it there until you manually dismiss it (thanks Christopher)
• Global can be dismissed with Escape (again)
• Fix for the manage models screen popping up on every launch if you only had WhisperKit models downloaded (thanks Anthony)
• Fix for the dictation feature asking for an OpenAI key even if you weren't using a prompt
• Fixed an issue where export styles were not displayed correctly in the quick export menu (thanks Steven)

9.5

Improved:

• Improved the preview of Markdown exports
• Removed the option to export a full transcript as Markdown as it was not working
• Fixed an issue where no spaces were shown in your dictation if you used a prompt (thanks Roel and Corey for reporting)
• The currently active prompt (or lack thereof) in dictation mode is now persisted across dictation sessions

9.4.1

Improved:

• Fixed an issue where the audio from the previous transcription would play when opening multiple files
• When using ChatGPT Prompts with dictation mode, the results are now typed as it comes in

9.4

New:

• Clicking the menubar icon will now open the Global window for a richer experience
• Support for Markdown (.md) exports

Improved:

• Dictation works in a lot more apps now and is more reliable
• The Dictation overlay is presented in the right place almost all the time now. If it doesn't let us know!
• When you add Whisper Transcription as a login item it will automatically be hidden when launched at startup of your Mac
• Batch files are now transcribed in the correct alphabetical order
• Export button is shown again when opening .whisper files
• Show model tags for downloaded and downloading model cells
• Buttons to quickly go to all models in the downloaded models screen
• Fix for activating WhisperKit models from the Manage Models screen
• Global now closes when you select “Open in main app”

9.3

Dictation improvements:

• If a blank dictation is detected, show an error in the overlay
• Rewritten typing engine, which is more robust (allowing streaming in future)
• Improved support for Safari and Firefox for Google Docs and ProtonMail
• Support for BBEdit 14
• Support for more electron apps

Fixed:

• Anthropic prompt never stops showing loading
• Export button is shown again when opening .whisper files
• Segment textfield: ⇧+return should always split the text instead of unfocusing
• Segment textfield: Allow splitting of line even when half the split is empty string
• Segment textfield: Can press Up key on empty line to go to previous segment
• Removed flag emojis

9.2

New:

• Added support for the new GPT-4o-mini model which is very cheap and still allows 128.000 tokens

9.1

New:

• The app is now localized into Dutch, more languages coming soon

Improved:

• Fixed some localization issues
• Models are now sorted in the right order in the model picker dropdown
• Global and menubar no longer save recordings to your preferred folder
• Improved the performance on the global find and replace settings screen if you have a lot of replacements
• Stopping global with Escape or the keyboard shortcut now correctly ends the recording
• When a transcription is empty this is now also shown on the segments page

9.0

New

• Dictation! Access high quality transcription anywhere on your Mac
• Added support for .m4b and .flac files

Improved

• When exporting srt subtitles with speakers, there is now a space after the “:”. (Thanks Marabel!)
• API keys are now hidden in settings by default and look nicer
• The model download page is a lot simpler and clearer
• The language selector in batch settings looks correct again
• Updated to the latest version of WhisperKit
• Updated ignore word list based on your feedback, thanks!
• Fixed an issue with Anthropic Claude 2.1
• Fixed an issue where microphone recordings could not be used with the OpenAI cloud transcription feature
• Global view is now always on top so you can keep using other apps while the transcription is active
• Global and menubar features now also work with cloud transcription

8.11

New

• You can now use the Cloud transcription feature for all parts of the app. Just select it from the model picker dropdown.
• Export translated transcripts from the export dropdown. When using the export dropdown with a translated transcript you can now choose which version of the transcript is exported.

Improved

• Fixed an issue where some old .whisper files could not be opened.
• The app shows a progress bar when loading models now.
• Decreased the size of whisper files with audio by converting the audio track of videos to m4a.
• Videos automatically show up in picture in picture mode after loading.

8.10

New

• You can now import and export your global replace list. This makes it easier to share common word replacements with friends or colleagues.

Improved

• Loading .whisper files with large video will now be 99% times faster since we first load the text and then later the media file!
• Find and replace settings now looks nicer and the filter works better
• When adding a new speaker the name textfield is auto selected to make your workflow faster
• Extra checks for validating your DeepL license key
• Fixed an issue where YouTube videos were not downloading if you had “Download video file” enabled

8.9.1

New

• Added support for the latest Anthropic model Claude 3.5 Sonnet
• You can now adjust the starting timestamp for the transcripts. Useful if you’re working with timecoded files. For example, you can now set the starting timestamp of the first segment to 01:00:00 and then all other timestamps will update accordingly. You can access this setting from the menubar > Transcript.

Improved

• When loading a large .whisper file you can now cancel the loading process
• Fixed some keyboard interactions and sound effects that were happening when they shouldn’t have
• Fixed some memory leaks
• Improved performance when switching between display modes
• Updated the ignore word list based on your feedback
• Changes in the segments view flow over to the AI view
• Hitting Shift-Return now creates a new segment. Before this was done with just Return, which will now commit the change in the textfield.
• Fixed a crash in system app audio recording
• The segments view animations are less messy when searching

8.8

Improved
• Fixed an issue where audio decoding could fail when using WhisperKit models. The app now falls back to an alternative audio decoding strategy.
• The AI features now use the correct token limits for each model

8.7

New
• Added a "Replace" button for replacing individual words alongside "Replace All".
• Select all segments up or down from the current selection using ⌘+⇧+⇧ or ⌘+⇧+⇧.
Improved
• Fixed: “System Audio Recording” and “Transcribe Podcast” sometimes didn’t display all tracks in the finished transcript.
• Fixed: Crash when starting Global transcription via ⌘+R with an ongoing transcription.
• Find and Replace now supports Undo (⌘+Z).
• Models causing crashes won't be automatically loaded on next launch.
• Corrected timestamp duration when backspacing to join two segments.

8.6

New
• Added emojis for each language so they’re easier to find
• Fast switching between translated languages in a transcript. Just tap the the flag to switch to another language.
Improved
• This update has a lot of design tweaks. Nicer buttons with prettier corners, cleaner header bar and playback bar
• Cleaned up the close buttons
• Added some more colors
• The url input bar shows an icon depending on what type of url you paste in

8.5

Improved
• The menubar app will only copy your transcript to the clipboard when it’s finished
• Clearer errors on the upgrade to pro screen if you don’t have an internet connection
• Improved handling of in app purchases when you are offline
• Fixed a crash when trying to add more than two speakers when not a Pro user (thanks Tom!)
• The language and quality badges are shown again in the header bar, oops!
• Added more words to the ignore list based on your feedback, keep submitting them please!
• More improvements to .ogg support

8.4

Improved
• Fixed an issue where the devices list was not updated correctly when a microphone was disconnected.
• System App Audio recordings are now automatically merged before being saved in your history.
• System App Audio playback now plays back both audio tracks correctly.
• Cleaned up the design for the System App Audio screen.
• When you close the app you are recording, you can continue it later.
• Improved an issue where some people were getting a “Runner not ready” error.
• You can now control more settings related to WhisperKit.
• The menubar icon will now show a different icon while transcription is happening.
• If you’re exporting a CSV with speaker names but a column doesn’t have speakers, it’s now empty instead of filled with a timestamp.
• Fixed an issue where .ogg files weren’t working for people (thanks Raymond!).
• Added the option to choose if you want to save only the audio or also the video file when making a whisper file for a video.
• Some small design tweaks.

8.3

New:
• You can now quickly toggle the “Translate to English” option from the Transcript menubar (⌘T)

Improved
• You can now open the find and replace bar from the menubar or with a keyboard shortcut (⌘⌃F)
• If you have the “Translate to English” setting enabled it will now be shown as a badge on the transcript view as a reminder
• Fixed an issue where the headerbar did not have the correct padding
• The language selector now shows languages you have not used before in a deeper menu to clean it up a bit
• You can now change the input language from the “Whisper” menubar item

8.2

New:
• Added a new quick export dropdown menu next to the Export button for even faster exporting

Improved
• Fixed some of the names for specific models
• Improved the Ignore Segments feature and made it only available for Pro users
• The language and used engine are now shown when opening .whisper files
• Added a warning before specialising WhisperKit models if your Mac is on low power mode

8.1

New:
• Ignore Segments. Commonly used filler segments can now automatically be removed to clean up your transcripts. And you can also add your own from Settings > Ignored Segments.
• Right click a segment on the Segments view to ignore it in the future
• Word Level Timestamps. You can now split up segments per word when using Whisper C++ models. Enable it from Settings > Advanced.

Improved
• Video files with capitalised file extensions now also show the videoplayer (thanks yuniancong)
• The timestamp for the first segment now starts at 00:00:001 for better compatibility with srt files
• WhisperKit is now disabled on Intel Macs and on Macs running Ventura

8.0

New Features:

• 📺 Video Player: Now, when transcribing video files, an inline video player is available! It can also be popped out into its own window. Subtitles display directly on the video, and translations appear as separate subtitles too.


• 🏎️WhisperKit Support: Choose different Whisper engines for your transcriptions. WhisperKit offers distilled models for speed, and transcriptions stream in real time. Enable WhisperKit in Settings > Advanced.

Improvements:
• If you have a character limit set, the app will not cut off words in the middle of a word.
• New menubar icon that doesn't conflict with the standard microphone icon.
• Quality and language selectors moved to the toolbar. Expand your window if they're not visible.
• Opening .whisper files is now possible while models load.
• Updated to the latest Whisper C++ engine, now with Flash Attention (activate in Settings > Advanced).
• Redesigned Manage Models screen for easier model selection. Feedback is welcome.
• Enhanced error handling for model downloads.
• "MS Teams Virtual Mic" excluded from microphone options as it's not an actual mic.
• Fixed a bug where invalid license error codes weren't displayed.
• Resolved a crash when non-pro users added more than two speakers.
• The Esc key won't close screens during active processes like recording or batch transcription.

Global:
• Keyboard shortcut modifiers now displayed in the UI (⌘+R etc).
• Improved button design.
• Fixed transcript copy errors.

YouTube:
• Faster YouTube downloads.
• Option to download only audio or video from YouTube.
• Downloads play in the mini-player.
• Choice of high or low video quality.

Cloud Transcription:
• The Cloud Transcription feature now only lists the languages that are supported (57) compared to the 100 that are supported locally
• Fixed the bug where m4a/mp4 files were being rejected even though they are supported
• All the file formats that the local transcription mode supports are now supported for Cloud as well

ChatGPT:
• Now always shows the network error if there is one.
• Support for latest models (latest GPT-4 Turbo, and GPT-4o!)

7.13

Improved:
• Fix phantom window being opened on Ventura when opening a file from Finder.
• Fix issue with nothing happening when opening a .whisper file from Finder on Ventura.
• Present error if there’s an issue when restoring a purchase.
• Remove “copy all” keyboard shortcut override (⌘+C) on Transcripts screen, so that you can still copy a text selection
• Commit pending textfield change when opening Export sheet, to ensure that an export contains the most recent changes.
• Add a note for Cantonese language import, stating it works best with the Large v3 model.
• Reset AI output when running a successive prompt.

7.12.1

• Small bug fixes

7.12

New:
• Added support for Anthropic Claude as the AI provider. Use the powerful Claude models to perform AI features on transcript with up to 200.000 tokens.
• Enhancements to the translation process during export. Now, on the export page, you can choose which language you wish to export to.
• Translations are now stored in your .whisper file.
• Laid the groundwork for making the app available in other languages.
Improved:
• Introduced a button to facilitate translating the app into your native language. It will remain accessible until we've recruited sufficient translators.
• Eliminated a flicker observed when initiating a new transcription.
• The translation button now displays the currently visible translated language for clearer language identification.
• Implemented comprehensive undo/redo functionality for translation entries.
• Bugfix on Undo when switching files
• "Combine Segments into Sentences" also supported for translations now.
• Improved the check for active Pro subscriptions. If you run into issues with your Pro license please contact us.

7.11

New
• Combine Segments into Sentences: Under "Transcript → Combine Segments into Sentences," you can now transform your fragmented segments into complete sentences, while preserving timestamps and metadata like favorites and speaker assignments.
• Audio Track Saving: When transcribing videos, now only the audio track is saved in the .whisper file, reducing file size and loading times.
• Exporting Segments: You can now choose to export segments with or without including milliseconds.
• Added compatibility for .aac audio files.
Improved
• Resolved an issue where .whisper files with an embedded .mov would not play back audio.
• Microphone Selection is saved: The app now remembers the last-used microphone selection correctly.
• Fixed sentence grouping in HTML and PDF exports.
• Cloud Transcription: Increased network timeout duration to accommodate longer transcriptions.
• Fixed a crash when using the "File → Export → Whisper" menu option.
• Media is added to history even if transcription is cancelled.
• DeepL API Key Error: Now shows an informative error if the DeepL API key is invalid or expired.
• Resolved a UI glitch in Batch Settings.
• Disabled the close button on the manage models screen while downloading.

7.10

New
• "Record App Audio" and "Transcribe Podcast" features now support playback of all recorded audio tracks.
• Additionally, there is an option to export a merged audio track combining all recorded tracks.
• ChatGPT: Introduced support for gpt-3.5-turbo-0125.
Improved
• Addressed a crash on Intel Macs during "Record App Audio" sessions. While we continue to investigate the root cause, recordings on Intel Macs may instead stop with an error within the first five minutes.
• Resolved a bug in macOS 14.4 where the main MacWhisper window disappeared upon opening a .whisper file from Finder.
• Provided a link to create an OpenAI API key.
• YouTube: Expanded support for audio streams, increasing the likelihood of successfully downloading from a YouTube URL.
• YouTube: Enabled downloading the same YouTube video twice without a “duplicate file” error.
• YouTube: Improved error messaging to clarify the cause of download failures.
• Added visual indicators for keyboard shortcuts to adjust playback speed (⌘+ to increase and ⌘- to decrease).
• Ensured that dismissing a Batch transcription properly resets it for future use.

7.9

New
• Added support for .ogg and .opus files
• You can now translate files into multiple languages
Improved
• Improved the translation feature. You can now choose to automatically translate the full transcript (with context) and the segments individually as well. Or you can choose to only translate the active display mode
• Removed the emoji on the podcast view
• You can change the colors again on the podcast view
• You can now open whisper files made with pro models even if you are not a pro user

7.8.1

New
• Cloud Transcriptions. You can now choose to use the OpenAI API version of Whisper to transcribe your files. This will run the transcription on their fast servers at the highest quality. Great for if you’re using an older Mac. Note that this requires an OpenAI API key, and will thus cost money. Files sent to OpenAI are no longer only stored locally on your Mac so be aware of this for private recordings.
• You can now choose which microphone is used in Global and the menubar app
• Added a tile on the homescreen to open the batch transcription feature
• Added keyboard shortcut hints in Global mode
• You can now adjust timestamps for segments. Click on the timestamp to adjust the start and end time by whole seconds. This is still very early so let us know what you would like to see us improve!
Improved
• You can now copy your entire transcript with a keyboard shortcut (⌘+C)
• When copying your transcript it will now take into account if you are in sentence or full transcript mode.
• When autosaving you will not see “Saving…” anymore in the header bar
• Audio playback stops when closing the transcription window
• Merging multiple segments into one now happens in the correct order (thanks Nathan!)

7.7.1

Fixed
• Fixed an issue that would prevent the app from saving

7.7

New
• Changes to .whisper files are now automatically saved when you go back to the main screen
• When you’re editing .whisper files, the app will autosave changes every 10 seconds
• You can now pause microphone recordings
• You can now merge multiple segments into one by selecting multiple and then selecting “Merge Segments” from the context window
• You can now ⌘+ tap on the textfield in a segment to select the segment itself instead of the textfield
Improved
• Saving and opening .whisper files is now 500% faster
• YouTube transcripts now use the video title as the filename
• YouTube transcripts should appear even faster
• When the whisper model can’t be loaded the alert will show the error code for easier debugging
• You can now change the “Show Timestamps” and “Large Font Size” settings from the View menu in your menubar
• Fixed an issue where the English only models could still use the last language used that was not English
• Improved the performance when showing large transcripts
• When you make changes to the segments they now show up in the ChatGPT display mode

7.6

New
• You can now choose which OpenAI model is used for the ChatGPT screen (GPT 3.5 or GPT 4)
Improved
• Cleaned up the some UI glitches in the batch screen
• Fixed an issue where the batch screen may not always appear
• Added an extra explanation to the Auto detect language button that explains how the language is determined
• Fix for custom model-loading crash, and strategy to prevent in future
• Added a retry button to the model downloader screen in case the list of models could not be downloaded
• If batch transcription fails the app shows a clearer error
• Small design tweaks on the System App Recording screen
• YouTube downloads work more reliably and should not error out

7.5

Improved
• YouTube transcriptions should download a lot faster now
• Improved transcription quality and performance on large audio files (30+ minutes)
• The back and forward skip buttons now both work in their correct direction
• Fixed an issue where the progress and scroll back buttons would overlap
• You will no longer see an error when you cancel a 'Save As...'
• Make it clearer that on first use of the recording feature you have to choose a folder to save the files to

7.4

New
• You can now drag in your own custom GGML models to use (Pro only). Use this with custom models trained on specific languages or datasets. You can download these from sites such as Hugging Face.
Improved
• When you add a new segment by hitting Return at the end of the line, the new segment will be automatically focused
• When the microphone is unplugged during a recording your recording is saved properly and you see an alert notifying you about what happened
• The milliseconds display when timestamps are visible now uses your Locale settings to determine if a period or comma should be used.
• When you scroll the segments page during playback, autoscroll will briefly be disabled and a button appears to enable it again.
• Fixes and improvements to playback performance on segments view.

7.3.1

New
• You can now rename the audio files in a System App Recording transcript before the transcription starts. This way you can tag your microphone audio as you, and for example the Teams audio as the name of a colleague
• You can now favorite or unfavorite a selected segment (or segments) by hitting the F key.
Improved
• Fixed an issue where files could not be saved if you did two transcriptions in the same minute
• Small tweaks to the transcription screen if no transcripts were generated
• The current selected segment is no longer unfocused if you assign a speaker to it with the keyboard shortcuts
• The global find and replace feature is now case-insensitive by default. You can still toggle this on if you prefer that behaviour.

7.3

New
• MDM deployments can now disable remote features such as translation and ChatGPT
• Audio files from system app recordings are now saved to your history for easy access
• Microphone recordings are saved to your history as well
• You can create an empty segment after the current one by pressing Return at the end of the textfield
• You can skip forwards and backwards in the audio by 5 seconds with the mediakeys on your keyboard
Improved
• Fixed an issue where opening a .whisper file could present an alert but still open the file correctly
• You can now save active .whisper files with Save, without being prompted to overwrite the file
• Fixed an issue where system audio recordings would stop without an error when putting your Mac to sleep
• Fixed an issue the system audio recording screen could not be closed
• Fixed an issue where the Pro status was not checked correctly which could lead to some users not being able to access a Pro feature
• Search highlights now appear in yellow for better legibility
• When editing a segment, it will stay active even if the highlighted segment that's currently playing changes
• You can transition from one segment to the next by pressing the left and right arrow keys at the start or end of a segment
• Improved then connection to OpenAI for the ChatGPT feature to prevent timeouts for long transcripts
• You can now change the playback speed by hitting Cmd - + and Cmd - -
• You can go back one step in playback speed by holding shift and pressing the speed button
• Disabled the copy button on the Export screen for binary files such as pdf and whisper to prevent crashes

7.2.1

• New
• Assign speakers directly with keyboard shortcuts (⌘+1 etc)
• Improved
• The app shows the detected language when using the Auto Detect language setting
• Fixed an issue where (1) was wrongly being added to an exported file even though there was no file with the same name in the directory
• Translated text now also shows as sentences if you enable the sentence mode in the Segments display mode.
• Fixed an issue where the export format wasn't displayed correctly in the batch export screen
• Available microphones in the picker are now shown faster
• When you disconnect your microphone during a recording, the recording is saved for transcription still
• After you delete a segment, the next one is selected for easier editing
• Batch files are now sorted alphabetically
• The model quality is now shown while a transcription is active

7.1

• New
• You can now add your own custom prompts to use with the ChatGPT feature
• Improved
• Improved the quality of longer transcriptions.
• Your last used export method is remembered
• You can group the full transcript export by sentences again even if you have not added any speakers
• Fixed export for docx in batch mode
• Batch exports will no longer overwrite files with the same name but will add (1) instead
• The prompts you use will be sorted by most recently used

7.0

• New
• ChatGPT integration! Add your own OpenAI API key and process your transcripts directly with ChatGPT. This is an early version, so I would love to hear your feedback! It requires you have access to GPT4-Turbo in this version.
• You can now use the new export styles in batch mode as well
• Improved
• You can now export again from the menubar, using the new styles
• The speaker paragraphs export option works again even if you have not added any speakers
• When batch exporting, the app will no longer overwrite existing files with the same name.
• You can now choose how you want to group the full transcript export (full, segments or sentences)
• System App Audio recordings will now stop if you close the app that you are recording

6.11

• New
• New and improved export screen. You're now able to customize what your exports look like in more detail. More improvements are coming over the next few weeks.
• Improved
• Fixed a crash when trying to overwrite a file during export
• You can now close the Global window by using the Escape key
• Added a nicer gradient to the top of the home screen
• Added some shine animations to the home buttons when hovering over them
• The Global view will no longer show transcripts created in the main app

6.10

• Improved
• When you enable Auto Start for Global mode, you are now able to start a new recording immediately when opening it again. Before it would not show the back button.
• Scroll indicators now are positioned correctly on the home screen
• Transcripts created in the main app are no longer shown in the Global window
• The settings window will no longer ask you to save you transcript before closing it

6.9.1

• Improved
• Add a button to show the sidebar in settings and added titles for each section

6.9

• Improved
• Redesigned the settings screen to provide more space for future features.

6.8

• Improved
• The app now stays open when you close the last window again

6.7.1

• New
• You can now adjust the Whisper temperature from Advanced settings
• Improved
• Set the max beam size to 5
• Gave some Whisper errors clearer descriptions

6.7

• New
• Added the option to use Beam Search instead of Greedy for improved transcription results. If you are running into duplicate segments, give it a try and let me know if it solves your problems.
• Improved
• Made some error messages clearer
• Fixed an issue where sometimes a file directory could not be opened
• Fixed an issue when splitting a segment

6.6

• Improved
• Improvements to the Global experience, thanks for the feedback
• The support buttons on the homescreen buttons work again

6.5

• New
• You can now open files and folders in MacWhisper directly from finder. Right click > Open with...
• Improved
• Removed the filename from the toolbar for now to not push the extra buttons to the more menu where they don't work

6.4

• New
• New Global mode! Access high quality transcription from anywhere with a keyboard shortcut. A spotlight like window will appear where you can immediatly start recording. The finished transcript can then be (auto) copied to your clipboard for easy pasting anywhere on your system.
• Press backspace to delete a selected segment in the Segments view
• You can now search through the Global Find & Replace list
• Improved
• Fixed padding and design issues on the System Audio screen
• The microphone recording screen now loads a lot faster
• Added a new experimental way to decode audio files that are giving problems. You can enable this from Settings > Advanced
• Transcriptions made in the menubar app no longer automatically show up in the main app

6.3

• New
• Added support for the new Large V3 model for even higher accuracy
• You can now change the language and model you want to use in the System Audio Recording screen and the Batch Transcription screen
• Improved
• Moved the progress bar into the header for a cleaner look
• The filename of the currently open file is now shown in the toolbar
• The time remaining and total progress is now shown during batch transcriptions
• Added a way to send Diagnostics Reports to us to help solve problems in the future
• The “Manage Models” button is now accessible again from the selector in the bottom left of the main screen
• The export format selection buttons are more easily tappable
• Fixed the counter in find and replace to be accurate
• Added the option to manage your Pro subscription

6.2

• New
• The progress bar now shows an estimated time remaining alongside the progress percentage
• You can now add a prompt to use for your transcription. This can help the app to better understand the context of the audio file you are transcribing. Examples of prompts would be: "this is a conversation between two English people" or "this is a conversation about rockets, words used are [ROCKET RELATED WORDS HERE]". You can find this feature under Settings > Advanced
• Add an option to hide milliseconds in the timestamp view on the segments page
• Added support for Undo and Redo in the segments view
• Improved:
• Extracted whisper files are removed from the temporary folder after they are opened
• Created wav files used during transcription are removed from the temporary folder when the transcription is finished
• It is now easier to open the audio files that are recorded during a System Audio recording.
• When searching in the transcript view, the page will scroll to the first occurence of the word and not just highlight it
• Added help text to the homescreen buttons to make it clearer what they do
• Added a button to open the Manage Models screen from the main page.
• On the Record System Audio screen the list of open apps is now updated live when you open or close apps
• You can now press the Return ⮐ key while you have a segment selected to edit the text
• You can hit the Escape key to unselect the text
• You can now navigate between selected segments with the arrow keys on your keyboard
• You can now select multiple segments and then hit ⌘+C to copy them as a whole to your clipboard
• Whisper files now open faster, especially the first one you open
• Redesigned the history screen to look nicer. More improvements to this are coming soon
• Fixed an issue on HTML export for batch transcriptions where the title was not correct
• Filter FaceTime (because its audio is not available for privacy reasons) and MacWhisper from the available apps to record on the System Audio screen.
• Fixed a small gap between segments when timestamps were enabled

6.1

• Fixed an issue where performance on Intel Macs was slow
• Fixed a design issue on Upgrade to Pro view

6.0.1

• New:
  • Metal support! The transcription process now runs using your GPU with the Metal framework. Especially on Apple Silicon Macs this leads to 2 to 3x speed improvements! Let us know if you run into anything related to this.
  • You can now play and pause audio playback by pressing the spacebar
  • You can use your media control buttons to control the audio
  • Added a fast way to export from the menu bar. File > Export...
  • MacWhisper will appear in Control Center when playing back audio
  • Added support for Cantonese
• Improved:
  • Fixed the Hebrew language setting not working correctly in some cases
  • m4v files can be opened again
  • Audio files with audio panned to either the left or the right will work properly now

5.7

• New:
  • Added support for notifications to remind you when a transcription has been finished.
  • You can now drag in (multiple) folders to perform batch transcriptions.
• Improved:
  • The quality of transcriptions should be better for certain files. If you were seeing repeated sentences, please let me know if this update fixes it.
  • Made it clearer that the language you select in the bottom left is the input language of the audio that you want to transcribe.
  • You should now be able to cancel an ongoing transcription without having to wait a long time.
  • Made the app 1MB smaller by removing some very large wallpapers that were only used in small sizes.
  • Show a "save confirmation" alert in more situations to prevent data loss.
  • Fixed an issue where the speed toggle didn't work for a very small number of people, let me know if it still happens for you.

5.4

• New:
• You can now add more files to the batch transcription window after it's been opened.
• Improved:
• You can now translate from the Segments view as well. Use this if you want to export your translated transcript.
• You can now split a segment by just pressing return. Hit shift-return to commit or click outside of the segment
• If you remove all text from a segment it will be deleted automatically
• While splitting segments the cursor will automatically move to the correct segment, making it easier to control with just your keyboard
• The "transcribe podcast" and "transcribe recording" buttons can no longer be tapped multiple times leading to strange behaviour in the app
• If you have enabled "play sound when finished" it will now also work on the menubar app
• The search/filter no longer uses the full url of the file but just the last part
• Increased history size from last 50 to 200

5.3.1

• Fixed an issue where transcripts would not complete if the Remove Duplicates features was turned off.

5.3

• New:
• Sometimes the transcription framework would return a lot of duplicate segments. This should no longer happen. You can disable this feature in Settings > Advanced.
• You can now set a maximum character limit for segments (useful if you want to adhere to the BBC subtitles size for example)
• You can now choose the location where recordings are saved
• You can now translate segments as well with DeepL
• Added support for all DeepL languages


• Improved:
• The app will now prompt you to make sure you really want to close a window or quit the app if it could lead to data loss
• Fixed typing glitch associated with committing a change
• Fixed an issue where colors were set incorrectly when switching between dark and light mode
• When searching you can now automatically scroll to the selected rows
• Fixed an issue where the segment highlighting would jitter while transcription was still being finished
• Improvements to splitting segments
• Fixes to selecting text in segments view

5.2

• Fixed an issue where sometimes the transcription would get stuck at 0%. Thanks for letting us know if you ran into this

5.1

• New:
• Split up segments! Press shift and return in the middle of a segment to split it up into a new line. More improvements coming to this area soon!
• Sentence view is back! On the transcripts page you can now click the sentence button in the top left to display your transcripts in a more structured formatted
• New save audio button to more easily export the audio file associated with a transcript. Click the waveform icon in the top left.
• You can now delete all occurrences of a segment by right clicking. Useful if there's some repeated sentences. We'll add more options to auto remove these in the next two weeks.
• Improved:
• Fixed an issue where the File menu did not show save or open buttons
• The pro upgrade screen can now be presented more consistently from the menubar app
• Searching through transcripts is now a looooot faster
• You can now open .whisper and audio/video files from each of the "open" flows in the app

5.0.1

• Menu bar app! Quickly dictate recordings from the menubar app and copy them into any textfield you need.
• View how long it took to transcribe the file after the transcription is finished
• Quickly open the original audio file for a (microphone) transcription by clicking the waveform icon in the bar at the top
• Added advanced settings to disable the confirmation alert when closing a transcript that has not been saved yet
• Loading .whisper files is faster now
• The progress bar is no longer shown when loading a .whisper file
• More consistent design for buttons in the header bar
• Files that are no longer available are automatically removed from your history
• The settings button now also works on newer operating systems
• Fixed an error when pasting a non-url into the url download bar
• You can now right click on the text as well in the segments view if you want to perform an action on that segment
• Fixed an issue where in system audio recording multiple recordings would stack
• Fixed an issue where you could not save a system audio recording (right now only your microphone audio is saved in the .whisper file)

4.6

• Added a button to unlock all features on the home screen if you're not using MacWhisper Pro

4.5

• If you click the back button from the transcription page without saving, an alert will pop up to ask if you want to save your transcript.
• The DeepL translation now takes into account punctuation (pretty silly we overlooked that) and works with more languages now.

4.4

• Fixed an issue where the microphone recording was not transcribed during System Audio Recordings.
• When "Play sound when finished" is enabled, the app will now only play the sound at the end of batch, podcast and system audio recordings and no longer for each file.

4.3

• Fixed an issue where new users couldn't see the language and model quality selectors. Sorry about that!

4.2

• New Home Screen design
• History now shows up to 50 of your last used files
• You can now search through your history
• The speed selector can now be long pressed to select a specific speed
• The used language is now presented in the top bar after a transcription is finished
• Added support buttons to each section which give you more info on how to get the most out of the features
• The system audio page now looks more in line with the rest of the app
• Find and replace does not show “0 matches” when you have not entered any text yet
• The translate button now shows even if you have not added your DeepL api key

4.1

• Fixed a crash when searching for words in a transcript
• You can now switch display modes while a file is being transcribed
• Speakers are now available on the top level menu when right clicking a segment instead of in a sub menu?
• You can toggle the auto scroll feature in the segments mode from the menu bar Transcripts > Toggle Auto Scroll
• The record feature will now remember the last input device you used and will default to that if it’s available
• Added a speaker limit of two for non Pro users

4.0.1

• The focus for version 4.0 was performance and speed! We spent weeks rewriting the segments view in the app so that it scrolls fast even for super long transcripts.
• You can now edit all text directly in the segments mode, without having to first click on a text label
• Right click on the background of a segment to favorite, add speakers or delete the segment
• You can now fill in any video or audio url to transcribe them directly, not just YouTube urls
• Double click a cell to start playback from it
• Improved scroll to segment performance while playing a track
• You can now save your audio recordings from the File menu
• Toggle timestamps on or off for the Segments view from within settings
• YouTube made some changes under the hood so some videos might not be able to be transcribed
• Fixed an issue where the textfield for adding names for podcast hosts would lose focus while you were typing

3.5.1

• You can now export transcript to a .whisper format from the export / batch screens as well. Useful for if you want to transcribe files and save them as a .whisper file which contains the audio AND the transcripts and edits you've made. You can also use File > Save as... to achieve the same thing
• YouTube downloads are significantly faster and now shows a progress indicator so you know how long a download will take.
• Added a view to compare the quality of the different transcription quality models. This will help you decide which model you need for your purposes.
• Added the option to save batch transcription exports to the same folder as the files you selected.
• A bunch of quality of life improvements and design tweaks
• The translation feature now supports more input languages such as Japanese and Russian
• The models should now be able to be downloaded more reliably if your network has restrictions (such as a VPN or work network)
• Fixed an issue where exports in paragraph view did not match the preview
• Fixed a lot of small stuff that were noticed by Harrison!

3.3.1

• This update fixes a bug where the app could crash when selecting a microphone in System Audio recordings. Sorry about that and thanks for letting me know so quickly so I could put out a fix!

3.3

• Improved performance when viewing large transcripts. Scrolling should be snappier now.
• Advanced Translation! You can now translate entire transcripts by using your own (free) DeepL API key. You will need a free (or Pro) DeepL API key, it's very easy to set up and you'll get 500.000 free translation characters every month. Be aware that when you translate your transcript, the content is sent to the DeepL servers for translating. This feature requires Whisper Pro. Right now you can translate into six different languages, but more are coming soon. Please send me any feedback on how you want to use it.
• You can now select which input device to use for your Microphone Recordings or System App Audio Recordings
• Made Find and Replace clearer

3.2

• Improved the recording screen experience.
• Your recording audio volume is now displayed to make it clearer that your microphone is picking up what you're saying
• Fixed an issue where audio you recorded could not be played back on the transcription screen
• If you have denied permission for the System App Audio Recording feature, the app will now redirect you to the settings when you click on the menu option
• WebVTT exports now display the speaker names in front of the transcript if you've added a speaker

3.0

• Starting with MacWhisper 3.0, new updates will only support macOS 13.0 (Ventura) and up. I had to update my test device to the new Sonoma beta, and it's very hard to support older versions of macOS while keeping my sanity. You can continue using up to 2.21 on Monterey for as long as you need. You can download old versions of the app from the Gumroad page.
• New display mode which shows the full transcript as one long piece of text. Thanks for all the requests!
• Made it easier to toggle between display modes
• You can now choose where to export batch transcription files to
• Batch export to DOTE file format
• Export to full transcript text file

2.20

• Podcast Transcriptions! Easily transcribe your podcast by providing audio files for each host and MacWhisper will automatically transcribe them, separating each speaker's dialogue. Please keep in mind that this feature is still in beta testing, so you may encounter some issues. This feature will be Pro only starting in a later release, and is only available on macOS Ventura and up.
• Save and load your transcriptions in a .whisper file format! You can now save transcribed files as .whisper files which you can easily share with others. They will include the audio file as well, so they can open them as if they made them themselves! Let me know if you run into anything!
• You can now play a sound to be notified when a transcription is finished
• Added a settings screen which you can access from the toolbar, menubar or by pressing "Cmd + ,". You'll find some common settings there that used to be in the toolbar.
• Improved the design of the batch settings screen.
• You can now access your recently used apps more easily for system app recording
• Your recently used languages are now shown at the top of the language picker list
• Greatly reduced memory usage when using different models in the same app run
• Fixed a bunch of small bugs here and there
• Editing segments performs a lot better now
• Show icons in the history

2.17

• Fixed an issue in System Audio Recording mode where the audio for the app recording would fail.

2.16

• Fixes an issue where the microphone recording during System Audio recording could keep recording after you finished.
• Microphone recordings are now saved with unique names to your Documents directory instead of as output.wav

2.15

• Up to 40% speed boost! MacWhisper can now use all the CPU cores on your Mac. For M1/M2 Pro/Max computers this should result in around 40% faster transcription!
• Added initial implementation for recording system audio from apps. This features is only available on Ventura because it uses APIs that are only available on Ventura. This feature will become for Pro users only in a later release. There will be bugs, so please report them to support@macwhisper.com.
• Rewrote the foundations for the model downloader so they should fix issues with downloads
• If a model download stops halfway through you can resume it later
• Models are now grouped based on if they're English only or Multilingual
• Added support to export to the DOTE transcription format
• New app icon!

2.13

• Manual speaker selection! You can now add speakers from the toolbar and then right click on single or multiple segments to add speakers. This is still very early and work in progress so please send me feedback :)
• Batch export now works properly on Monterey
• You can now export to multiple formats at once when performing batch transcription
• Global Find & Replace can be accessed from everywhere
• Hopefully fixed an issue where the cursor would jump to the end of the segment when editing the text
• You can now adjust the text size

2.12

• Batch Transcription! Drag and drop multiple files on MacWhisper to transcribe and export them one after another. Great if you need to transcribe a large number of audio files at the highest quality. You can just leave your Mac running overnight and wake up to fresh transcripts. This feature is available for MacWhisper Pro users only.
• Added keyboard shortcuts to quickly open a file or start a recording (Cmd+O and Cmd+R) from the start screen
• Some small design tweaks here and there

2.11

• Transcription will now continue at full speed even if you run MacWhisper in the background!
• Made it clearer that you can not close the model downloader screen until you've downloaded at least one model or if you're downloading a model.
• Added buttons to open the Finder location for the downloaded models.
• Global Find and Replace. Add words or phrases to be automatically replaced in new transcripts. This can be helpful for accents or names. Note: Right now the system replaces the text wherever it's found (even within other words) so, for example, replacing “you” will also replace the same letters in the word “your.” Access it from the settings icon in the toolbar.
• You can now change the display mode for the transcript by clicking the display mode button in the toolbar
• In Reader mode you can switch between showing the whole transcript as one chunk, or split up by sentences. The copy button will adjust based on your current mode.
• The home screen now provides quick access to your three last used audio files (for now they will need to be re-transcribed each time, working on a way to save completed transcripts so you can continue working on them)
• Clicking on a segment will now no longer play from that location but instead will let you edit. You can play a segment by clicking the play button on the right side
• While editing a segment the text will no longer under/overlap with the buttons on the right side
• You can now export to HTML, as in, the button works
• You can now export to PDF as well

2.10

• Fixed an issue in 2.9 since that update was accidentally sandboxed (normally meant for the Mac App Store). If you downloaded 2.9 you will probably have to manually update to 2.10 by downloading it directly from https://macwhisper-site.vercel.app/releases/MacWhisper.zip. Before you update you should delete all the models that were downloaded (again) in version 2.9 as they are saved in a different directory and would otherwise take up space on your Mac. Sorry about this!

2.9

• You can now transcribe YouTube videos by pasting the url. This feature is only available for MacWhisper Pro users and only on Ventura, but you can test to see if it works if you're a free user as well :). Videos are downloaded to your documents folder for now, but please send me feedback on how well / not well this works for you!
• HTML Export! Export your transcript into an HTML page. This version is very early and needs some design love, but I'm not great with HTML and CSS so bear with me here :)
• If you copy the transcript from the toolbar or reader view it will now be exported as individual sentences instead of one big chunk
• Fixed an issue where the reader view could not be opened on Monterey
• Fixed an issue where export was not working on Monterey

2.8

• Fixed an issue where the app would crash on Monterey

2.7

• Added a new export preview screen where you can see what the output file will look like

2.6

• Favourite segments are now highlighted on the slider bar at the bottom
• Find and replace. You can now find and replace words across your transcript. Note that currently it will replace all occurrences of the string you're replacing, also if it's part of a larger word. Please send me feedback if you run into anything (Ventura only for now)
• The reader mode now splits up the transcript in sentences for easier, well, reading!

2.5

• Added a button in the toolbar that notifies when a new version is available

2.4

• Fixed an issue where the app would sometimes randomly crash when transcribing files while it would work on the next try with no issues. Thanks a lot for sending the crash reports!

2.3

• When you record audio with your microphone, the app will now show "New Recording" as the title of the file instead of an empty space
• You can now change the playback speed of the audio recording. Play your audio at 0.5x up to 3x speeds by toggling with the button in the bottom right of the playback bar.
• After you finish a microphone recording the app won't go back to the home screen until transcriptions are displayed.
• The file formats that are presented on the landing screen are no longer spoken through Voice Over.

2.2

• Fixed an issue where the microphone could not be used.

2.1

• Downloaded Whisper Models are now saved in the Application Support/MacWhisper/models directory and they're excluded from backups.

2.0

• The app is now very small! You will have to download the different quality levels manually, but they will persists across updates. This will make it a lot easier to handle updates in the future :)
• The app can now automatically update itself without you having to download it again from the website
• You can now favorite individual segments. This will be useful in a later version where you can save and load .whisper files
• The scrub bar now shows the segment text while scrubbing so you can more easily find specific parts of a transcript
• Click on a segment to play
• Drag and Drop Voice Memos directly from the Voice Memos app into MacWhisper
• Edit a segment by clicking on the edit button
• Design tweaks to make the app nicer to look at
• Fixed a crash where 8 bit mp3 files weren't able to be transcribed
• Added a warning for users with 8GB of RAM to inform them that the higher quality transcription levels might not work on their device.
• Improved support for dropping .m4a files
• You can show your transcript in Compact Mode which hides the timestamps