Facts

Dictation software: facts and numbers

342 facts, each traced to the exact words of its source.

A 1997 Boston Globe columnist wrote that older programs, such as DragonDictate, forced users to pause between each spoken word.Simson L. Garfinkel, The Boston Globe, 1997
A 2018 report says WhatsApp's CarPlay app let you tap its icon to start a new message by dictation, but did not let you browse your conversation list.9to5Mac, 9to5Mac, 2018
A 2024 news photo caption describes Draft One as an AI tool that uses bodycam audio to generate police reports.Kerry Klein, KQED (with KVPR, Fresno), 2025
A 2025 preprint on older Dutch adults talking to a geriatric chatbot found that generic multilingual speech recognition models outperformed models fine-tuned for older Dutch speakers.Bram van Dijk et al., arXiv (forthcoming in the HCINLP workshop proceedings at EMNLP), 2025
A December 2020 archived copy of Google's Docs help page says voice typing is only available in Chrome browsers.Google Docs Editors Help (archived by the University of North Texas Web Archive), 2020
A February 2015 Dragon Medical sheet for eClinicalWorks users listed as a key benefit the ability to dictate anywhere in eClinicalWorks.eClinicalWorks and Nuance Communications, 2015
A health system's Epic tips say Dragon Medical One is the tool for dictating into a patient's chart on a desktop or tablet in Hyperspace.Connect Care (Epic) tips blog, a Canadian health system, 2020
A reseller states that Nuance and Microsoft did not create a Dragon Home version 16.Voice Power NZ Ltd, n.d.
A reseller's notice says Dragon Home is no longer available for sale.Voice Power NZ Ltd, n.d.
A UK reseller sells the Dragon Professional V16 Educational Volume License for schools, colleges and universities, and prices it per user.Speak-IT Solutions Ltd, n.d.
A US reseller describes Dragon Professional Anywhere as cloud-hosted speech recognition for the enterprise.StartStop, n.d.
A user wrote in a 2017 Microsoft Q&A post that the Failed to Launch UI Automation Server message began after a Windows 10 update, with Dragon NaturallySpeaking 12.Microsoft Q&A, 2017
According to the launch report, clinicians can use voice commands to change the structure of the AI-drafted note, such as formatting current conditions as a bulleted list.Healthcare Dive, 2026
According to the VoiceInk README, buying a license supports development and gives access to automatic updates, priority support and upcoming features.GitHub (VoiceInk repository), 2026
After ending the Mac line, Nuance said customers could still get telephone support for up to 90 days from activation in North America.The Register, 2018
57 Firefox Quantum versionAn older Nuance help page for the Dragon Professional web extension, labelled version 4.3, lists Mozilla Firefox Quantum version 57 or later as a supported browser. The Dragon 16 pages do not list Firefox.Nuance Communications, Nuance Communications, older Dragon Professional help, undated
An older Nuance help page says the Dragon Professional Web Extension is not applicable to Microsoft Edge, so you cannot click web page objects by voice in Edge.Nuance Communications, Nuance Communications, older Dragon Professional help, undated
An older Nuance install procedure, labelled version 4.3, says Firefox shows a prompt to install the Dragon Professional web extension the first time you open Firefox after Dragon is installed.Nuance Communications, Nuance Communications, older Dragon Professional help, undated
Apple says Journal stores your entries in iCloud, where they are backed up automatically.Apple, Apple App Store, 2026
Apple says that once Speak selection is turned on in Read & Speak settings, the keyboard shortcut makes the Mac start speaking, and its default key combination is Option-Esc.Apple Inc., Apple Support, Mac User Guide, n.d.
Apple says that when you dictate in a search field on an iPhone, the dictated text may be sent to the search engine to process the search.Apple Inc., Apple Support, iPhone User Guide, n.d.
Apple says that with iOS 15 and later, Siri can announce your incoming messages in the car.Apple Inc., Apple Support, 2026
Apple says that, on a Mac, voice inputs and transcripts for general text Dictation are processed on your device and not sent to Siri servers, and that you can check this in Keyboard settings.Apple Inc., Apple Support, Mac User Guide, n.d.
Apple says the Journal app lets you record your voice directly into a journal entry and have it transcribed automatically.Apple, Apple App Store, 2026
Apple says you can dictate text of any length on a Mac without a timeout.Apple Inc., Apple Support, Mac User Guide, n.d.
3 SiriKit intentsApple's CarPlay guide says a communication app that provides text messaging must support three SiriKit intents: send a message, request a list of messages, and modify the attributes of a message.Apple Inc., Apple Developer, 2026
Apple's CarPlay guide says communication apps in CarPlay provide their features through SiriKit Messaging, VoIP calling, or both.Apple Inc., Apple Developer, 2026
Apple's CarPlay guide says iOS automatically disables the keyboard and shortens lists when the car indicates it should do so.Apple Inc., Apple Developer, 2026
Apple's CarPlay guide says that in many cars the keyboard is not available at all while driving.Apple Inc., Apple Developer, 2026
Apple's Mac Help says you can check whether general text Dictation is processed on your device and not sent to Siri servers.Apple Inc., Apple Support, macOS User Guide, n.d.
Aqua Voice says that for users with Privacy Mode turned off, it may store transcript data on its servers to improve the product.Aqua Voice (vendor website), 2026
1,000 free wordsAqua Voice's free plan includes 1,000 free words and its Avalon transcription model.Aqua Voice, Aqua Voice (vendor website), 2026
1,000 wordsAqua Voice's pricing page says its Free plan includes 1,000 free words and the Avalon transcription model.Aqua Voice, Aqua Voice (vendor website), 2026
10 US dollars per month, billed monthlyAqua Voice's Pro plan costs $10 per month when billed month to month.Aqua Voice, Aqua Voice (vendor website), 2026
8 US dollars per month, billed annuallyAqua Voice's Pro plan costs $8 per month when billed annually.Aqua Voice, Aqua Voice (vendor website), 2026
AskJAN lists Dragon Naturally Speaking as an accessibility solution for people who cannot use a keyboard, and for students with physical challenges.Job Accommodation Network (AskJAN), n.d.
150 US dollarsAt its 2016 launch, Dragon Professional Individual was offered for $150 to users upgrading from Dragon Professional Individual version 14 and from Dragon Professional versions 12 and higher.MacTech, 2016
12.7 minutesAt the entry speeds in the phone study, typing 1,000 words on the built-in keyboard takes about 12.7 minutes longer than speaking them, before any editing.Calculated from other facts
180 US dollars per yearAt the monthly list price, one Wispr Flow Pro seat costs $180 over a year.Calculated from other facts
29.4 monthsAt the monthly price, Superwhisper's lifetime license costs the same as about 29 months of the monthly Pro plan.Calculated from other facts
285.7 words per dayAveraged over the week, the free plan's 2,000 desktop words per week comes to about 286 dictated words per day.Calculated from other facts
BigHand lists digital dictation and speech recognition among its solutions, aimed at accelerating dictation and transcription.BigHand, n.d.
Buzz describes itself as transcribing and translating audio offline on a personal computer.Chidi Williams, GitHub (Buzz repository), 2026
685.99 US dollarsCDW lists a one-user Dragon Professional (v. 16) license at $685.99 on the day we checked.CDW, n.d.
5 trial launchesDragon 16.1 asks you to activate on first start. If you skip activation, you can launch it in Trial mode five times before you must activate.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Dragon Copilot combines and extends Dragon Medical One and DAX Copilot.Microsoft, 2025
11 NaturallySpeaking version for PCDragon Dictate for Mac 2.0, released in September 2010, used the engine from Dragon NaturallySpeaking 11 for the PC.Kathleen Savino, Speech Technology magazine, 2010
Dragon NaturallySpeaking 12 became available to purchase as a download on August 3, 2012.Police1, 2012
99.99 US dollarsDragon NaturallySpeaking 12 was available for pre-order starting at $99.99 through Nuance's website and its reseller partners.Police1, 2012
4.8 versionDragon Professional 16 needs Microsoft .NET Framework version 4.8 or later, which Nuance says you must download from Microsoft if it is not installed.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Dragon Professional 16 says the web extension's features are not available in Chrome Metro mode, Chrome Incognito mode or Edge InPrivate mode.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
3 monitorsDragon Professional 16 supports MouseGrid commands on up to three monitors.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
300 US dollarsDragon Professional Individual will be available for download in the U.S., UK and Australia for $300 on September 1, 2016.MacTech, 2016
9 numbered areasDragon's MouseGrid divides your screen or active window into nine numbered areas, and you move the pointer by saying a number.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
DragonDictate was originally developed by Dragon Systems for Microsoft Windows and was later replaced by Dragon NaturallySpeaking for Windows.Wikipedia, n.d.
eClinicalWorks says its Scribe tool helps physicians save minutes per patient encounter when it is used with voice-recognition software for dictation.eClinicalWorks, 2024
eClinicalWorks says Sunoh, its ambient listening product, is fully integrated within the eClinicalWorks record.eClinicalWorks, 2024
Epic's Haiku (smartphones) and Canto (iPads) apps support Nuance Dragon dictation directly into progress notes started in the mobile app.Connect Care (Epic) tips blog, a Canadian health system, 2020
7 ratersFinal-year veterinary students and clinicians rated the transcripts and summaries, seven raters in total.Takuya Yogo, Frontiers in Veterinary Science, 2025
For the Open ASR Leaderboard's normalized transcripts, punctuation and casing are removed, along with disfluencies such as fillers, repetitions and repairs, before word error rate is computed.Hugging Face and collaborators, arXiv, 2025
Handy offers a Parakeet V3 model that its README describes as CPU-optimized with automatic language detection.CJ Pais, GitHub, 2026
Handy, a free open-source speech-to-text app, is listed as working on Windows, macOS and Linux.CJ Pais, GitHub, 2026
Handy's README describes its use as pressing a shortcut, speaking, and having your words appear in any text field.CJ Pais, GitHub, 2026
Handy's README lists Whisper models in four sizes, Small, Medium, Turbo and Large, and says they use GPU acceleration when a GPU is available.CJ Pais, GitHub, 2026
Handy's README says that transcription happens on your own computer without sending any information to the cloud.CJ Pais, GitHub, 2026
Handy's README states that the project is released under the MIT License.CJ Pais, GitHub, 2026
Healthcare Dive reported on February 5, 2026 that Epic had rolled out an AI tool that drafts clinical notes.Healthcare Dive, 2026
Heidi Desktop, the vendor says, lets a user dictate anywhere on the screen, capture telehealth audio from the call, and skip a second login.Heidi Health, n.d.
Heidi Dictate lets a user hold one key and speak, and Heidi writes the finished clinical text into the record, inbox or any form.Heidi Health, n.d.
90+ languagesHeidi's Dictate product page says the tool is ready from a user's first visit and works in more than 90 languages.Heidi Health, n.d.
Heidi's pricing page lists unlimited transcription among the Scribe features in its plan comparison.Heidi Health, n.d.
70,000 clinicians (vendor claim)Heidi's vendor page says it powers the largest AI scribe procurement in NHS history, covering 70,000 clinicians, 15 NHS trusts and over 1,200 GP practices.Heidi Health, n.d.
HEX describes itself as local-first voice dictation for Apple silicon Macs running macOS 15 or newer.Christopher Langton, GitHub (anomalyco organization), 2026
14.1 to 15.2 percentIn a 2000 peer-reviewed study of physicians dictating medical reports, Dragon NaturallySpeaking Medical Suite 3.0 had a vocabulary recognition error rate of 14.1 to 15.2 percent.Eric G Devine et al., Journal of the American Medical Informatics Association, 2000
In a 2008 longitudinal study on court recordings, the word error rate of a speech recognizer rose gradually as the age of elderly speakers increased.Ravichander Vipperla et al., Interspeech 2008 (ISCA), 2008
In a 2022 post, the whisper.cpp maintainer said the tiny.en Whisper model can run the real-time examples on a Raspberry Pi 4 Model B using 3 CPU threads.Georgi Gerganov, GitHub, ggml-org/whisper.cpp, 2022
10.3 percent word error rateIn a 2025 9to5Mac test of one 7-minute-31-second episode, Apple's speech API had a word error rate of 10.3 percent, against 1.5 percent for Whisper Large V3 Turbo.Marcus Mendes, 9to5Mac, 2025
In a 2025 study of speech recognition models on children's speech, a child's age and the number of words in the audio had the highest impact on accuracy, followed by background noise and pronunciation ability.Vishwanath Pratap Singh et al., arXiv, 2025
In August 2017, Lexacom said it was the only digital dictation provider with full integration with every UK clinical system, naming Microtest, Emis, SystmOne and Vision.Building Better Healthcare, 2017
In December 2013, Nuance said the latest PowerScribe 360 version was built using the same technology as Dragon Medical 12.Sandip Sen, Speech Technology News, 2013
In Dragon 15, Nuance tells you to say Spell That when a recognition error occurs, which opens the Spelling window.Nuance Communications, Nuance Communications, Dragon 15 help for PC, 2018
In Dragon 15, saying Correct That opens the Correction menu for the last text you dictated.Nuance Communications, Nuance Communications, Dragon 15 help for PC, 2018
In Dragon 15, Scratch That, Delete That or Undo that deletes the last text you dictated.Nuance Communications, Nuance Communications, Dragon 15 help for PC, 2018
In Dragon 15, the default Correction hot key is the minus sign key on the numeric keypad, and pressing it opens the Correction menu.Nuance Communications, Nuance Communications, Dragon 15 help for PC, 2018
In Dragon Professional 16, Chrome shows a prompt to install the web extension the first time you open Chrome after Dragon is installed.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
In Dragon Professional 16, Microsoft Edge shows a prompt to install the web extension the first time you open Edge after Dragon is installed.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
In Dragon Professional 16, Nuance says you say Go to close the MouseGrid, or say another number to move the pointer again.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
In Dragon's MouseGrid, you close the grid by saying Cancel, and Undo That reverses the last MouseGrid action.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
In May 2015, Nuance said PowerScribe 360 Reporting version 3.0 incorporated the Nuance Dragon Medical speech recognition engine.Speech Technology News, 2015
In November 2001, a court-approved sale of Lernout and Hauspie's speech assets included the Dragon NaturallySpeaking line, which was to go to ScanSoft.Gilbane, 2001
In October 2025, California's governor signed a law requiring police to disclose how they use generative artificial intelligence, the first law of its kind in the state.Kerry Klein, KQED (with KVPR, Fresno), 2025
In standalone mode, Dragon Professional 16 backs up your user profile automatically every 5 saves by default.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
7.0 to 9.1 percentIn the 2000 study, IBM ViaVoice 98 had the lowest mean vocabulary error rate, 7.0 to 9.1 percent.Eric G Devine et al., Journal of the American Medical Informatics Association, 2000
13.4 to 15.1 percentIn the 2000 study, the L&H Voice Xpress software had a mean vocabulary error rate of 13.4 to 15.1 percent.Eric G Devine et al., Journal of the American Medical Informatics Association, 2000
1.75 percent mean word error rateIn the 2025 neurosurgery pilot, the modified Whisper-based model had a mean word error rate of 1.75 percent, counting all errors.Benjamin S. Hopkins et al., Neurosurgical Focus, 2025
In the Dragon 16.1 install wizard, you enter the product serial number before you select your region and installation folder.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
In the Dragon Professional 16 release notes, Nuance says that after the Dragon Web extensions for Edge or Chrome are installed, dictation still works but you can no longer use Dragon commands.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
In the HyPoradise benchmark, a large language model can correct words that are missing from the recognizer's N-best list, because it generates text rather than only choosing among candidates.Chen Chen et al., arXiv (NeurIPS 2023 dataset track), 2023
54 percentIn the North East and North Cumbria ICB's October 2025 response, 188 of 348 GP practices (54%) used TPP SystmOne, against 160 (46%) on EMIS Web.NHS North East and North Cumbria Integrated Care Board, 2025
0.080 p valueIn the pilot, the difference in mean word error rate across all errors was not statistically significant.Benjamin S. Hopkins et al., Neurosurgical Focus, 2025
1.5 percent word error rateIn the same 2025 9to5Mac test, Whisper Large V3 Turbo had a word error rate of 1.5 percent on the episode, the lowest of the three engines the author tested.Marcus Mendes, 9to5Mac, 2025
1.54 percent mean word error rateIn the same pilot, Dragon Medical One had a mean word error rate of 1.54 percent, counting all errors.Benjamin S. Hopkins et al., Neurosurgical Focus, 2025
139.99 pounds sterlingIn the UK, Dragon NaturallySpeaking 13 Premium was available now from Nuance for £139.99, while the Home version, at £79.99, was due in August.Expert Reviews, 2014
In the veterinary study, the iPhone's built-in speech recognition gave transcripts that outperformed Whisper on readability, technical accuracy and clinical flow.Takuya Yogo, Frontiers in Veterinary Science, 2025
J-Dictate combines the screen reading of JAWS for Windows, made by Freedom Scientific, with the voice recognition of Dragon NaturallySpeaking from Nuance.Job Accommodation Network, Job Accommodation Network (JAN), AskJAN.org, n.d.
J-Dictate is used only for dictation, not for command and control of the computer.Job Accommodation Network, Job Accommodation Network (JAN), AskJAN.org, n.d.
J-Say is a third-party product that combines Dragon NaturallySpeaking professional from Nuance and JAWS for Windows, so a blind user can control the computer by voice.Job Accommodation Network, Job Accommodation Network (JAN), AskJAN.org, n.d.
KQED reports that the new California law requires written disclosures and supporting materials whenever departments use AI to generate police reports.Kerry Klein, KQED (with KVPR, Fresno), 2025
MacWhisper says it processes sensitive content locally, without data ever leaving the Mac.MacWhisper, MacWhisper (vendor website), 2026
Microsoft announced Dragon Copilot on March 3, 2025 as the healthcare industry's first unified voice AI assistant.Microsoft, 2025
Microsoft announced on Friday, March 4, 2022 that it had completed its acquisition of Nuance Communications Inc.Microsoft, 2022
5 English localesMicrosoft lists five English locales that Voice Access currently supports.Microsoft, Microsoft Learn, 2026
Microsoft says Dictation in Word works in Microsoft 365 and needs a reliable internet connection.Microsoft, 2026
Microsoft says that to uninstall a program in Windows 10 or Windows 11, you go to Start, then Settings, then Apps, then Installed apps.Microsoft Support, n.d.
Microsoft says the first-time setup of Voice Access needs an internet connection.Microsoft Support, n.d.
Microsoft says the initial setup of Voice access in Windows needs an internet connection.Microsoft Support, 2026
22H2 Windows 11 versionMicrosoft says Voice Access is available for Windows 11, version 22H2 or later.Microsoft, Microsoft Learn, 2026
Microsoft says Voice access is available in Windows 11, version 22H2 and later.Microsoft Support, 2026
Microsoft says Windows speech recognition is no longer being developed and is being replaced by Voice Access.Microsoft, Microsoft Learn, 2026
Microsoft's help for PowerScribe One says its cloud speech recognition needs no local voice profiles to train or maintain.Microsoft Support, 2026
Microsoft's quick start guide for Dragon Copilot in Epic says to open Epic Haiku or Canto on a mobile device, select a patient, select DAX Copilot at the bottom of the screen and start the ambient recording.Microsoft Support, n.d.
480 US dollars per yearMilner lists the Dragon Professional Anywhere annual subscription at $480 per year on a one-year term.Milner, n.d.
40 US dollars per monthMilner lists the Dragon Professional Anywhere monthly subscription at $40 per month on a one-year term.Milner, n.d.
NHS England expects GPs, ICBs and acute trusts to use its ambient voice technology registry as part of their necessary due diligence checks before buying.HTN Health Tech News, 2026
Nuance bought MacSpeech in February 2010. MacSpeech's own product, Dictate, was based on the Dragon product line.Kathleen Savino, Speech Technology magazine, 2010
Nuance Communications, Inc. was formerly ScanSoft, Inc.Wikipedia, n.d.
Nuance defines a Dragon user profile as a set of files that Dragon uses to process your speech, which stores information such as the type of microphone you use and your custom words.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance lists .NET 4.8, .NET 8.x (x86), ASP.NET Core 8.x (x86) and Microsoft WebView2 as the runtimes for Dragon Professional Anywhere 2026.1.Nuance Healthcare, 2026
Nuance lists Windows 10, Windows 11 and Windows Server 2016, 2019, 2022 and 2025 as operating systems for Dragon Professional Anywhere 2026.1.Nuance Healthcare, 2026
Nuance lists Windows 11 (64-bit), Windows 10 (32-bit and 64-bit) and Windows Server 2016, 2019 and 2022 as supported operating systems for Dragon Professional 16.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance recommends exporting your custom words and commands before you upgrade to Dragon 16.1, as a safeguard.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance said in 2017 that Dragon Law Enforcement lets officers dictate reports from their patrol cars while keeping their attention on their surroundings.Officer.com (reprinted Nuance press release), 2017
Nuance said it had decided to discontinue Dragon Professional Individual for Mac.The Register, 2018
22 October 2018 date updates endedNuance said it would stop providing updates for Dragon Professional Individual for Mac after 22 October 2018.Dictate.com.au, Dictate.com.au (news and blog, reseller), 2018
24 monthsNuance says a Dragon Professional Anywhere release that is more than 24 months old will no longer function, because of advances in its cloud infrastructure.Nuance Healthcare, 2026
Nuance says a USB microphone should be connected directly to the computer's USB port, not through a USB extension or hub.Nuance Communications, n.d.
Nuance says administrator rights are not required to create a user profile or use Dragon after installation.Nuance Communications, Inc., Nuance Communications, 2023
Nuance says Apple AirPods are not officially supported for dictation with Dragon 16.1.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance says Dragon Anywhere Mobile lets you create, edit and format documents of any length.Nuance Communications, n.d.
Nuance says Dragon Law Enforcement allows central management and customizations, including terms, street names and other words unique to an agency.Officer.com (reprinted Nuance press release), 2017
12 GBNuance says Dragon Professional 16 needs 12 GB of free hard disk space, including the downloaded installation files.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says Dragon Professional v16 is suited to two kinds of work: front-end applications, which turn live speech into text, and back-end applications, which transcribe audio files recorded earlier.Nuance Communications, n.d.
Nuance says Dragon Professional v16 uses a speech engine powered by Nuance Deep Learning technology.Nuance Communications, n.d.
Nuance says Dragon supports automatic upgrades from versions 14, 15 and 16 without an uninstall, and that it keeps your custom words and commands.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance says Dragon upgrades versions 14, 15 and 16 to version 16.1 automatically, so an owner of one of those versions does not need to uninstall it first.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance says Dragon's Read That reads editable text such as a Microsoft Word document or an email message.Nuance Communications, 2021
Nuance says earlier Dragon versions called the Read That feature Text-to-speech.Nuance Communications, 2021
Nuance says Read That is different from Playback, because Dragon does not record your dictation and save it when it reads text aloud.Nuance Communications, 2021
Nuance says that after an uninstall you can remove leftover user profiles yourself: in File Explorer, go to Nuance\NS16\Users and select the profiles to delete.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says that after an upgrade to version 16, Dragon moves a standalone user profile from the ProgramData folder to the LocalAppdata folder, and that a renamed profile left in ProgramData must be deleted from that folder.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says that Dragon Anywhere users who could not synchronize with version 16.0 can synchronize again, because the problem was resolved in version 16.1.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance says that if Dragon consistently misrecognizes words or phrases, you should consider training Dragon.Nuance Communications, Nuance Communications, Dragon 15 help for PC, 2018
Nuance says that if Dragon fails to move your user profile during an upgrade to version 16, it renames the profile.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says that if you bought the Dragon 16 download, the serial number is in the confirmation email from Nuance, in both the CSV file and the HTML file attached to it.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says that if you get no response when you speak, the first check is that Dragon's microphone is on, as opposed to off or asleep.Nuance Communications, n.d.
5 launchesNuance says that if you skip activation when you first start Dragon 16, you can launch it in Trial mode five times before you must activate it.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says that if you upgrade from Dragon Professional or Dragon Legal to Dragon Law Enforcement version 16, your user profiles are not updated automatically.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says that if you upgrade to Dragon 16.1 from Dragon 13 or earlier, you must export your words and commands, uninstall your current version, install 16.1 and create a new user profile when prompted.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance says that in standalone mode, Dragon upgrades your user profiles when you open them after an upgrade.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance says that to test your Windows sound system, you should go to the Windows Control Panel and click Sound.Nuance Communications, n.d.
Nuance says that user profiles stay on your local drive after you uninstall Dragon, and that you can remove them yourself if you want to.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says that when Nuance Management Center mode is on, Dragon creates dynamic user profiles in a folder under the user's AppData\Local\Nuance\NS16\Users path.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says that when you install Dragon 16.1, the installation process removes the previous version from your computer.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance says that when you upgrade Dragon, the serial number of your existing Dragon installation is required.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says that where a Dragon edition has Nuance Management Center (NMS) mode enabled, the user needs an internet connection to the Nuance Management Center server and a Nuance Management Center login, and an administrator provides the account and license.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance says the microphone check error can happen when the microphone is disconnected or when the sound card or audio device driver has a problem.Nuance Communications, n.d.
Nuance says users upgrading from Dragon 12 or earlier must follow a manual procedure, not the automatic upgrade used for versions 13 to 15.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance says volume licensing and the Nuance Management Center let organizations deploy, manage and share custom words, commands and auto-texts across users.Nuance Communications, n.d.
Nuance says you can easily upgrade between different versions of the same edition, for example from Dragon Professional Group version 15.x to Dragon Professional version 16.1.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance says you can have only a single Dragon user profile per region, and creates one for each region you dictate in.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance tells you to run Dragon's sound system check by saying "Check Audio", or through the DragonBar menus to Microphone and then Check Microphone.Nuance Communications, n.d.
Nuance warns that if an administrator creates a user profile while logged in as an administrator, that profile is not accessible to users with restricted privileges.Nuance Communications, Inc., Nuance Communications, 2023
September 1, 2018Nuance withdrew Dragon Premium 13 from the market effective September 1, 2018.Voice Power NZ Ltd, 2018
100 new featuresNuance's 2012 launch release said Dragon NaturallySpeaking 12 came with more than 100 new features and enhancements.Police1, 2012
24 percentNuance's August 2016 announcement of Dragon v15 said its deep learning technology delivers upwards of 24 percent greater accuracy.MacTech, 2016
Nuance's Dragon 16 help says you can back up your user profile manually from Settings, then Profile, then Back up User Profile, and that you can export it.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance's Dragon 16.1 instructions assume the computer where you install Dragon has an internet connection.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance's Dragon 16.1 known issues say that on some laptops, creating a user with a USB microphone can show a timeout error on the Check Microphone dialog.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
4 GBNuance's Dragon 16.1 release notes give 4 GB of RAM as the minimum for all operating systems.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance's Dragon 16.1 release notes say dictation needs an audio input device capable of 16-bit recording.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance's Dragon 16.1 release notes say the earlier problem with Dragon commands after installing the web extensions is fixed, and that all commands work after installing them.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance's Dragon for Mac help says that after you deactivate the license from the Help menu and confirm, Dragon confirms the license is deactivated.Nuance Communications, Nuance Communications, Dragon for Mac help, n.d.
250 computersNuance's Dragon K-12 School License Program is a single-site program for K-12 schools that lets a school install Dragon Professional on up to 250 school-owned or school-leased computers at one location.Nuance Communications, n.d.
Nuance's Dragon Professional 16 help has a Get trained page that offers short interactive training modules on key aspects of Dragon.Nuance Communications, 2023
4 GBNuance's Dragon Professional 16 release notes give 4 GB of RAM as the minimum on every supported Windows version.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance's Dragon Professional 16 release notes list a known issue: Dragon does not recognize dictation or commands when you use the Citrix environment in Generic Mode.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance's Dragon Professional 16 release notes say Dragon upgrades versions 13, 14 and 15 to version 16 automatically.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance's Dragon Professional group client guide for version 16 lists a /Deactivate switch for natspeak.exe, which deactivates the license for the current Dragon client.Nuance Communications, Inc., Nuance Communications, 2023
Nuance's Dragon Professional product page for Windows describes the product and then offers a Contact us link, with no price shown in the text of the page when we checked it on October 16, 2026.Nuance Communications, n.d.
Nuance's education page says its Open License Program is a public volume license program for qualified K-12 and HED organizations, and that it offers significant education discounts. The page gives no percentage or price.Nuance Communications, n.d.
Nuance's help for Dragon Professional 16 says the user profiles are in the Nuance\NS16\Users folder inside the local AppData folder of the user.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance's help for Dragon says its Read That feature lets Dragon read editable text aloud in a computer voice.Nuance Communications, 2021
Nuance's install steps for Dragon Professional 16.1 start with running the Dragon.exe installer by double-clicking it; the installer then downloads and installs the required files.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Nuance's microphone tips say not to use a handheld microphone if you dictate for long periods.Nuance Communications, Nuance Communications, Dragon for PC help, 2021
Nuance's microphone tips say that when you first start with Dragon, a headset is the best choice.Nuance Communications, Nuance Communications, Dragon for PC help, 2021
15 percentNuance's own sheet for Dragon Professional Individual v15 says accuracy improved by up to 15 percent out of the box, without training.Nuance Communications, 2016
August 31, 2026 end-of-life dateNuance's product life cycle table gives August 31, 2026 as the end-of-life date for Dragon Professional Anywhere 2021.1, a cloud-hosted release. The same date appears for several older Anywhere releases.Nuance Healthcare, 2026
Nuance's release notes for Dragon Professional Group 15.1 list version 13 and version 14 as the versions you can upgrade from, and version 12 is not on that list.Nuance Communications, Nuance Communications, n.d.
Nuance's release notes list support for the Microsoft Teams Home and Work editions as a new feature of Dragon Professional 16.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance's release notes list Windows 11 support, Office 2021 support and Windows Server 2022 support as new in Dragon Professional 16, along with MouseGrid on up to three monitors.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
90 daysNuance's statement said telephone support for discontinued Dragon Professional Individual for Mac would last up to 90 days from activation in North America.The Register, 2018
99 percentNuance's store page says Dragon Professional v16 lets employees dictate documents 3 times faster than typing, with up to 99 percent recognition accuracy.Nuance Communications, n.d.
Nuance's training guide for Dragon Professional Anywhere says the product's cloud-hosted technology keeps users on the latest version of Dragon.Nuance Communications, 2021
Nuance's US store menu lists Dragon Legal, Dragon Law Enforcement, Dragon Professional Anywhere and Dragon Legal Anywhere as products. Dragon Home and Dragon Premium are not in the menu.Nuance Communications, n.d.
Nuance's workaround for the Citrix Generic Mode issue is to run Citrix in optimized mode.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
Nuance's written statement, quoted in October 2018, said it had decided to discontinue the Dragon Professional Individual for Mac line.The Register, 2018
NVIDIA says its models are designed and optimized for NVIDIA GPU-accelerated systems.NVIDIA, Hugging Face, 2025
NVIDIA says the parakeet-tdt-0.6b-v3 model extends language support from English to 25 European languages.NVIDIA, Hugging Face, 2025
NVIDIA states that use of the parakeet-tdt-0.6b-v3 model is governed by the CC-BY-4.0 license.NVIDIA, Hugging Face, 2025
600 million parametersNVIDIA's parakeet-tdt-0.6b-v3 is a multilingual speech recognition model with 600 million parameters.NVIDIA, Hugging Face, 2025
4 GB memory required on Windows 11Olympus lists the memory requirement for ODMS R7 as 2 GB on Windows 10 and 4 GB on Windows 11, with server versions needing 4 GB or more.OM Digital Solutions, Professional Audio Support, n.d.
Olympus says the software key you supply decides which version of the ODMS R7 standalone launcher installs, for the dictation or transcription module.OM Digital Solutions, Professional Audio Support, n.d.
Olympus shows how to configure the Wi-Fi of its DS-9500 dictation recorder through ODMS R7 software.OM Digital Solutions, Professional Audio Support, n.d.
OM System lists the single license for the ODMS R8 dictation module as a one-time cost with no subscription, and as compatible with Windows operating systems.OM Digital Solutions (OM System), 2026
OM System says its ODMS R8 dictation software is compatible with both OM System and Olympus handheld dictation devices.OM Digital Solutions (OM System), 2026
On 16 October 2026 the Buzz releases page marked v1.4.5 as the Latest release.Chidi Williams, GitHub (Buzz repository), 2026
On 16 October 2026 the Handy releases page marked v0.9.8 as the Latest release.CJ Pais, GitHub (Handy repository), 2026
On 16 October 2026 the OpenWhispr releases page marked version 1.10.3 as the Latest release.GitHub (OpenWhispr repository), 2026
On 16 October 2026 the VoiceInk releases page marked v2.21 as the Latest release.GitHub (VoiceInk repository), 2026
On Superwhisper's Free plan, dictation uses local Whisper models without AI processing, and the company says nothing dictated leaves the device.Superwhisper, Superwhisper documentation, 2026
1,500 words per weekOn the free plan of Wispr Flow on iOS, dictation keeps going but slows down at 1,000 words in the weekly window, and a card blocks new dictations at 1,500 words.Wispr Flow Help Center, 2026
One Superwhisper Pro license works on unlimited Macs and Windows PCs, plus an iPhone and an Android phone.Superwhisper, Superwhisper documentation, 2026
OpenWhispr describes itself as a cross-platform dictation app for macOS, Windows and Linux.GitHub (OpenWhispr repository), 2026
OpenWhispr's README lists Whisper and NVIDIA Parakeet among the local speech models that run offline.GitHub (OpenWhispr repository), 2026
OpenWhispr's README states that it has no data collection and no telemetry, and that it is fully open source.GitHub (OpenWhispr repository), 2026
January 1, 2019Owners of Dragon Premium 13 kept a perpetual license, but Nuance stopped providing support or updates for it after January 1, 2019.Voice Power NZ Ltd, 2018
Philips describes its SpeechLive Health AI Assistant as a tool that converts clinical conversations into structured notes.Philips SpeechLive, 2026
78.00 US dollars per yearPhilips lists the SpeechExec Basic Dictate module, sold as a two-year subscription, at a recommended retail price of $78.00 per year in the US.Philips Dictation (US), 2026
Philips states that its VoiceTracer Speech Recognition Software works only with Philips VoiceTracer audio recorders.Philips Dictation (US), 2026
Philips' page for the VoiceTracer Speech Recognition Software DVT2805 is marked as a product that is no longer available.Philips Dictation (US), 2026
Philips' SpeechExec Basic page asks for a software serial number before it offers the installation file.Philips Dictation (US), 2026
-1.7% change in time-in-note versus control (not significant)Physicians who used Microsoft DAX Copilot showed no statistically significant change in time-in-note compared with the control group.Lukac PJ et al., NEJM AI, 2025
9.5% reduction in time-in-note versus controlPhysicians who used Nabla cut their time-in-note by 9.5 percent compared with the control group, the only arm with a statistically significant reduction.Lukac PJ et al., NEJM AI, 2025
Researchers describe word error rate as a standard measure of the discrepancy between machine and human transcriptions.Allison Koenecke et al., Proceedings of the National Academy of Sciences, 2020
Samsung's Australian support page calls Bixby its AI assistant, built into Samsung phones and smart watches, and says you can command it by voice, text or tapping gestures.Samsung Australia Support, 2026
Samsung's India support page says that with the keyboard open, you press and hold the Bixby key to start Bixby dictation.Samsung India Support, 2024
Samsung's US support page says that once Bixby is set up, you can wake it by pressing and holding the Bixby button or the Side button.Samsung US Support, 2026
7.08 US dollars per monthSpread over twelve months, Superwhisper's annual plan works out to about $7.08 per month.Calculated from other facts
660 US dollars per yearStartStop lists the Dragon Professional Anywhere annual subscription at $660 per year on a one-year term.StartStop, n.d.
55 US dollars per monthStartStop lists the Dragon Professional Anywhere monthly subscription at $55 per month on a one-year term.StartStop, n.d.
8.49 US dollars per monthSuperwhisper lists its Pro plan at $8.49 per month, billed monthly.Superwhisper, Superwhisper documentation, 2026
Superwhisper Pro adds cloud models, AI processing, file transcription and more to the Free plan.Superwhisper, Superwhisper documentation, 2026
Superwhisper says Pro adds cloud models, AI processing, file transcription and more on top of the free plan.Superwhisper, Superwhisper documentation, 2026
249.99 US dollars, paid onceSuperwhisper sells a lifetime Pro license for $249.99 once, which covers unlimited Macs and Windows PCs plus iPhone and Android.Superwhisper, Superwhisper documentation, 2026
Superwhisper's documentation recommends using your own API keys for organizations that already have HIPAA business associate agreements or financial data compliance agreements with an AI provider.Superwhisper, Superwhisper (vendor documentation), 2026
Superwhisper's documentation says that a dictation can skip its language model stage, and that its Voice to Text mode outputs raw transcribed text with no language model involved.Superwhisper, Superwhisper (vendor documentation), 2026
Superwhisper's free plan gives unlimited dictation using local Whisper models, with up to two modes, custom vocabulary and text replacements.Superwhisper, Superwhisper documentation, 2026
Superwhisper's home page lists unlimited use of Whisper models among the features of its free plan.Superwhisper, 2026
Superwhisper's home page says the app works offline, so you can transcribe without a Wi-Fi connection.Superwhisper, 2026
249.99 US dollars, one timeSuperwhisper's Pro plan is listed at $249.99 as a one-time lifetime purchase.Superwhisper, Superwhisper documentation, 2026
8.49 US dollars per monthSuperwhisper's Pro plan is listed at $8.49 per month on a monthly plan.Superwhisper, Superwhisper documentation, 2026
84.99 US dollars per yearSuperwhisper's Pro plan is listed at $84.99 per year on an annual plan.Superwhisper, Superwhisper documentation, 2026
30,000 veterinary professionalsTalkatoo's home page says it is trusted by more than 30,000 veterinary professionals worldwide.Talkatoo, n.d.
Talon's home page offers downloads for macOS, Windows and Linux (X11).Talon Voice, n.d.
Talon's home page says it is for hands-free input for writing, coding, browsing and gaming.Talon Voice, n.d.
TechRadar's 2023 review concludes that Dragon Professional 16 is an effective speech-to-text program for everyday use.Kieron Allen, TechRadar, 2023
TechRadar's 2023 review lists three drawbacks of Dragon Professional 16: it consumes computing resources, it is expensive, and it has a steep learning curve.Kieron Allen, TechRadar, 2023
699 US dollarsTechRadar's 2023 review of Dragon Professional 16 says it is sold for a one-time fee of $699 for each license.Kieron Allen, TechRadar, 2023
The 2007 PowerScribe deployment at Yale-New Haven was planned to integrate with the hospital's Radiology Information System and Picture Archiving and Communications System, giving direct access from within the PACS.Speech Technology Magazine, 2007
The 2017 Microsoft Q&A poster reported that Dragon still did not open after User Account Control was turned back on.Microsoft Q&A, 2017
The 2017 Microsoft Q&A poster said that Dragon opened only after they ran a repair from their installation media.Microsoft Q&A, 2017
The 9to5Mac author who ran the 2025 test wrote that Apple's new transcription API is faster than Whisper, before testing how accurate it was.Marcus Mendes, 9to5Mac, 2025
The anomalyco/hex LICENSE file is the MIT License.Christopher Langton, GitHub (anomalyco organization), 2026
1.4.5 last version for Intel MacsThe Buzz README says the Mac version now requires Apple silicon, and that version 1.4.5 is the last one that supports Intel Macs.Chidi Williams, GitHub (Buzz repository), 2026
The COPPA Rule requires operators of covered services to give parents direct notice and obtain verifiable parental consent before they collect, use or disclose personal information from children under 13.Federal Trade Commission, Federal Register (US Government Publishing Office), 2025
The Dragon 16 group client guide lists a /FindAppData switch for natspeak.exe, which opens the AppData directory for the Dragon client.Nuance Communications, Inc., Nuance Communications, 2023
2 browsersThe Dragon Professional 16 help lists two browsers for the Dragon Professional Web Extension, Google Chrome and Microsoft Edge, each in its latest version.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
February 28, 2023The encyclopedia article's infobox gives February 28, 2023 as the stable release date of Dragon version 16.Wikipedia, n.d.
August 16, 2016The encyclopedia's version table gives August 16, 2016 as the release date of Dragon version 15, Dragon Professional Individual.Wikipedia, n.d.
August 2010The encyclopedia's version table gives August 2010 as the release date of Dragon version 11, in the Home, Premium, Professional and Legal editions.Wikipedia, n.d.
August 2014The encyclopedia's version table gives August 2014 as the release date of Dragon version 13, in the Home, Premium, Professional and Legal editions.Wikipedia, n.d.
September 2015The encyclopedia's version table gives September 2015 as the release date of Dragon version 14, Professional (individual and Group).Wikipedia, n.d.
The footer of Nuance's US store page for Dragon Professional carries two copyright notices, one for Microsoft and one for Nuance Communications, Inc.Nuance Communications, n.d.
20 Scribe uses per weekThe free plan of Willow Voice includes an unlimited weaker speech-to-text model and 20 Scribe uses per week.Willow Voice (vendor website), 2026
The FTC's amended children's privacy rule adds biometric identifiers that can be used to recognize a person to the definition of personal information.Federal Trade Commission, Federal Register (US Government Publishing Office), 2025
The Handy project describes itself as a free, open source speech-to-text application that works completely offline.CJ Pais, GitHub, 2026
The HEX README says its Linux beta has a smaller feature set and requires a source build or Nix.Christopher Langton, GitHub (anomalyco organization), 2026
The HEX README says its menu lets you switch between downloaded models, for example English with Parakeet and French with Whisper.Christopher Langton, GitHub (anomalyco organization), 2026
The HEX README says that when post-processing with OpenCode is turned on in Modes, the app sends the transcript, the foreground application and the Brave website hostname to the configured model provider.Christopher Langton, GitHub (anomalyco organization), 2026
The HEX README says you download a local speech model during setup, and that ordinary dictation does not need OpenCode or a cloud transcription service.Christopher Langton, GitHub (anomalyco organization), 2026
The HEX README says you hold a shortcut, speak and release it, and that HEX transcribes on your Mac and pastes the result into the app you are using.Christopher Langton, GitHub (anomalyco organization), 2026
9.11 percent word error rateThe Kid-Whisper study reports reducing the word error rate of Whisper-Small on the MyST children's test set from 13.93% to 9.11%.Ahmed Adel Attia et al., arXiv (eess.AS), revised May 2024, 2023
1550 million parametersThe largest Whisper model, large, has 1,550 million parameters and needs about 10 GB of memory.OpenAI, GitHub (OpenAI repository), 2023
The LICENSE file in the VoiceInk repository is a short GPL header that points to the complete license text online.GitHub (VoiceInk repository), 2026
The license file in the whisper.cpp repository is the MIT License, which allows free use, copying and modification of the code.ggml-org, GitHub, 2026
The license file of Buzz, an offline Whisper transcription app, is the MIT License, with the copyright line Copyright (c) 2022 Chidi Williams.Chidi Williams, GitHub, 2022
2.2 GHzThe minimum processor for Dragon Professional 16 is a 2.2 GHz Intel dual-core or equivalent AMD processor, and faster processors give faster performance.Nuance Communications, Nuance Communications, Dragon Professional 16 help, 2023
The ODMS R7 release notes say that Windows 11 and Windows Server 2022 became supported, alongside Office 2021 and Azure Virtual Desktop.OM Digital Solutions, Professional Audio Support, n.d.
1.04 times the annual subscriptionThe one-user Dragon Professional v16 license at $685.99 costs about the same as one year of the Dragon Professional Anywhere subscription at $660.Calculated from other facts
2.29 times the 2016 priceThe one-user Dragon Professional v16 license listed at $685.99 is about 2.29 times the $300 launch download price of Dragon Professional Individual v15 in 2016. The two are different editions and years, and the 2016 price is not adjusted for inflation.Calculated from other facts
86 systemsThe Open ASR Leaderboard compares 86 open-source and proprietary speech recognition systems across 12 datasets.Hugging Face and collaborators, arXiv, 2025
The Open ASR Leaderboard paper says token-and-duration transducer (TDT) decoders offer superior speed (RTFx), which makes them better suited for long-form and batched processing.Hugging Face and collaborators, arXiv, 2025
The Open Source Definition requires that the source code be the preferred form in which a programmer would modify the program.Open Source Initiative, Open Source Initiative, 2024
The Open Source Initiative's license page lists the GNU General Public License version 3.0 with an approval date of September 5, 2007.Free Software Foundation, Open Source Initiative, 2007
The OpenAI Whisper repository states that its code and model weights use the MIT License.OpenAI, GitHub (OpenAI repository), 2023
The OpenWhispr LICENSE file is the MIT License, which allows free use, copying and modification.OpenWhispr Team, GitHub (OpenWhispr repository), 2024
The original Swift Hex repository was archived by its owner on 14 October 2026 and is now read-only.Christopher Langton, GitHub (kitlangton), 2026
3 physiciansThe pilot study had three independent physicians dictate and record the neurosurgical procedures.Benjamin S. Hopkins et al., Neurosurgical Focus, 2025
The preamble of the GNU General Public License version 3.0 says that when it speaks of free software, it means freedom, not price.Free Software Foundation, Open Source Initiative, 2007
The preamble of the GNU General Public License version 3.0 says the license is meant to give you the freedom to distribute copies of free software, and that you may charge for them if you wish.Free Software Foundation, Open Source Initiative, 2007
39 million parametersThe smallest Whisper model, tiny, has 39 million parameters and needs about 1 GB of memory.OpenAI, GitHub (OpenAI repository), 2023
The study defines word error rate with the total number of words in the correct transcript as its denominator.Allison Koenecke et al., Proceedings of the National Academy of Sciences, 2020
The study used Nuance's Dragon Medical One as its commercial comparison.Benjamin S. Hopkins et al., Neurosurgical Focus, 2025
The study's AI system was a modified speech-to-text model built on the backbone architecture of OpenAI's Whisper.Benjamin S. Hopkins et al., Neurosurgical Focus, 2025
238 physiciansThe UCLA ambient scribe trial assigned 238 outpatient physicians across 14 specialties to two AI scribes or to usual care.Lukac PJ et al., NEJM AI, 2025
The University of York says the Plantronics Savi 400 Series wireless headset comes with a USB dongle, and that the headset will not work without it.University of York IT Services, University of York, n.d.
The University of York says VoicePower, a Nuance partner for Dragon, can recommend the Plantronics Blackwire C510 as a wired headset for Dragon.University of York IT Services, University of York, n.d.
4 devicesThe University of York's IT services describe the Dragon Professional Individual licence as one that lets the user install Dragon on up to 4 devices for their own use.University of York IT Services, University of York, n.d.
The University of York's IT Services says Individual Education licences are per device, which means the software is installed on a single device.University of York IT Services, University of York, n.d.
5 clinical case discussionsThe veterinary study that compared iPhone speech recognition with Whisper recorded five clinical case discussions.Takuya Yogo, Frontiers in Veterinary Science, 2025
The VetSkribe App Store listing says the app syncs with most veterinary practice management systems.Apple App Store, n.d.
The VoiceInk README says the latest version is available with a free trial from tryvoiceink.com.GitHub (VoiceInk repository), 2026
The VoiceInk README says you can build the app yourself from its source, and that the compiled version adds automatic updates, priority support and funding for development.GitHub (VoiceInk repository), 2026
The What's new section of Nuance's Dragon Professional 16.1 release notes lists LibreOffice 64-bit support, followed by bug fixes, security enhancements and stability work. It lists no other new feature.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
The Whisper repository describes Whisper as a general-purpose speech recognition model trained on a large dataset of diverse audio.OpenAI, GitHub (OpenAI repository), 2023
244 million parametersThe Whisper small model has 244 million parameters and needs about 2 GB of memory.OpenAI, GitHub (OpenAI repository), 2023
To get the University of York's Dragon Professional Individual Education licence at the education discount, a qualified staff member or student must submit a valid University of York email address to Nuance to obtain a serial number.University of York IT Services, University of York, n.d.
VoiceInk's README claims that 100% offline processing keeps the user's data on the device.GitHub (VoiceInk repository), 2026
VoiceInk's README credits whisper.cpp for running OpenAI's Whisper model and FluidAudio for its Parakeet model implementation.GitHub (VoiceInk repository), 2026
VoiceInk's README describes it as a native macOS application that transcribes speech to text almost instantly.GitHub (VoiceInk repository), 2026
VoiceInk's README says the project is licensed under the GNU General Public License v3.0.GitHub (VoiceInk repository), 2026
25 US dollars, paid once, one MacVoiceInk's Solo license costs $25 once, for one Mac, with lifetime access and no monthly bill.VoiceInk (vendor website), 2026
VoiceInk's website says it uses local AI models to transcribe what you say.VoiceInk (vendor website), 2026
When a Superwhisper user supplies their own API keys, the app acts as a client of that user's account with the AI provider, and the provider agreement, not Superwhisper's terms, governs how the data is handled.Superwhisper, Superwhisper (vendor documentation), 2026
1.34 percent mean word error rateWhen linguistic errors were excluded, Dragon Medical One had a mean word error rate of 1.34 percent.Benjamin S. Hopkins et al., Neurosurgical Focus, 2025
0.50 percent mean word error rateWhen linguistic errors were excluded, the Whisper-based model had a mean word error rate of 0.50 percent.Benjamin S. Hopkins et al., Neurosurgical Focus, 2025
When Nuance Management Center mode is enabled, Dragon needs an internet connection to reach the NMC server.Nuance Communications, Nuance Communications, Dragon Professional 16.1 help, n.d.
Wispr Flow lists dictation on Mac, Windows, iOS and Android.Wispr Flow, Wispr Flow (vendor website), 2026
15 US dollars per user per monthWispr Flow lists its Pro plan for individuals and teams at $15 per user per month.Wispr Flow, Wispr Flow (vendor website), 2026
Wispr Flow says it deletes your audio as soon as it has been transcribed, unless you have turned on its Dictation cloud storage setting.Wispr Flow (vendor website), 2026
2,000 words per week on desktopWispr Flow's free plan allows 2,000 dictated words per week on desktop and 1,000 per week on mobile, according to its pricing page.Wispr Flow, Wispr Flow (vendor website), 2026
Wispr Flow's Free plan includes dictation on Mac, Windows, iOS and Android, and the page lists it at $0 per month.Wispr Flow, Wispr Flow (vendor website), 2026
Wispr Flow's help center says the desktop weekly count is a combined total of desktop and mobile dictation.Wispr Flow Help Center, 2026
Wispr Flow's home page says the app automatically removes filler words, adds punctuation and formats your writing as you dictate.Wispr Flow, Wispr Flow (vendor website), 2026
Wispr Flow's home page says the app understands what you mean, not just what you say, so you can ramble, pause or change your mind mid-sentence.Wispr Flow, Wispr Flow (vendor website), 2026
0 US dollars per monthWispr Flow's pricing page lists its Free plan at $0 per month, including speech to text in any app and 100+ languages.Wispr Flow, Wispr Flow (vendor website), 2026
15 US dollars per user per monthWispr Flow's pricing page lists its Pro plan for individuals and teams at $15 per user per month.Wispr Flow, Wispr Flow (vendor website), 2026
Wispr Flow's privacy page has an Improve the model for everyone setting, which lets the company use audio, transcripts and edits to improve its models. The company says that turning it off means nothing you say is used to train its models.Wispr Flow (vendor website), 2026
Wispr Flow's Pro plan includes everything in the Free plan plus unlimited dictations.Wispr Flow, Wispr Flow (vendor website), 2026
12 US dollars per user per monthWispr Flow's Pro plan is listed at $12 per user per month, and its Growth plan at $18 per user per month.Wispr Flow, Wispr Flow (vendor website), 2026
2.7 timesWithout linguistic errors, Dragon Medical One's mean word error rate was 2.7 times the Whisper-based model's rate in the pilot.Calculated from other facts
365,000 reports processed each yearYale-New Haven Hospital's 118-member radiology department planned to use PowerScribe to cut turnaround for more than 365,000 reports a year.Speech Technology Magazine, 2007